Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Synthesis of thermoelectric magnesium-silicide pastes for 3D printing, electrospinning and low-pressure spray

Synthesis of thermoelectric magnesium-silicide pastes for 3D printing, electrospinning and... In this work, eco-friendly magnesium-silicide (Mg Si) semiconducting (n-type) thermoelectric pastes for building compo- nents concerning energy-harvesting devices through 3D printing, spray and electrospinning were synthetized and tested for the first time. The Mg Si fine powders were obtained through the combination of ball milling and thermal annealing under Ar atmosphere. While the latter process was crucial for obtaining the desired Mg Si phase, the ball milling was indispensable for homogenizing and reducing the grain size of the powders. The synthetized Mg Si powders exhibited a large Seebeck coef- ficient of ~ 487 µV/K and were blended with a polymeric solution in different mass ratios to adjust the paste viscosity to the different requirements of 3D printing, electrospinning and low-pressure spray. The materials produced in every single stage of the paste synthesis were characterized by a variety of techniques that unequivocally prove their viability for producing thermoelectric parts and components. These can certainly trigger further research and development in green thermoelectric generators (TEGs) capable of adopting any form or shape with enhanced thermoelectric properties. These green TEGs are meant to compete with common toxic materials such as Bi Te , PbTe and CoSb that have Seebeck coefficients in the range 2 3 of ~ 290–700 μV/K, similar to that of the produced Mg Si powders and lower than that of 3D printed bulk Mg Si pieces, 2 2 measured to be ~ 4866 μV/K. Also, their measured thermal conductivities proved to be significantly lower (~ 0.2 W/mK) than that reported for Mg Si (≥ 4 W/mK). However, it is herein demonstrated that such thermoelectric properties are not stable over time. Pressureless sintering proved to be indispensable, but dic ffi ultly achievable by long thermal annealing (even above 32 h) in inert atmosphere at 400 °C, at least for bulk Mg Si pieces constituted by a mean grain size of 2–3 μm. Hence, for overcoming this sintering challenge and become the silicide’s extrusion viable in the production of bulk thermoelectric parts, alternative pressureless sintering methods will have to be further explored. Keywords Mg Si-based thermoelectric materials · Mechanical alloying · 3D printing · Electrospinning · Spray * A. C. Marques Departamento de Física e CICECO, Instituto de Materiais de acl.marques@fct.unl.pt Aveiro, Universidade de Aveiro, 3810-193 Aveiro, Portugal C2TN, Instituto Superior Técnico, Universidade de Lisboa, CENIMAT/I3N, Departamento de Ciência dos Materiais, Campus Tecnológico e Nuclear, Estrada Nacional 10, Faculdade de Ciências e Tecnologia, Universidade Nova de 2695-066 Bobadela, LRS, Portugal Lisboa, 2829-516 Caparica, Portugal IPFN-IST/UL, Instituto de Plasmas e Fusão Nuclear, Instituto Dipartimento di Scienza Applicata e Tecnologia, Politecnico Superior Técnico, Universidade de Lisboa, Estrada Nacional di Torino, Corso Duca degli Abruzzi 24, 10129 Turin, Italy 10, 2695-066 Bobadela, Portugal I3N/Departamento de Física e CICECO, Instituto de Materiais de Aveiro, Universidade de Aveiro, 3810-193 Aveiro, Portugal Vol.:(0123456789) 1 3 21 Page 2 of 10 Materials for Renewable and Sustainable Energy (2019) 8:21 and lower energy conversion efficiencies of the devices. A Introduction new approach is herein devised to overcome the challenge of shape: it consists in the production of Mg Si powders The increasing energy demand worldwide has been driv- through a simple and cost-effective process (relying on the ing the search for new, clean, renewable and sustain- combination of ball milling with thermal annealing), for able energy sources. Solar, wind and hydropower energy subsequent formulation of thermoelectric pastes suitable sources are expected to fulfill future energy needs and for 3D printing, electrospinning and spray technologies. replace energy sources based on fossil fuels. However, The major problems with the Mg Si powder synthesis and currently, these still assure about 90% of the world’s elec- paste formulation are related to the high reactivity of Si tricity generation with low operating efficiency (30–40%) and Mg powders with oxygen, demanding the use of an and large annual waste of heat to the environment (15 inert atmosphere, e.g. a glove box filled with Ar, and lim- TW) [1]. Such large amount of wasted heat can be directly its the selection of solvents and polymers to oxygen-free converted into electricity by solid-state generators based compounds. One should note that the need for develop- on the thermoelectric (TE) materials, using the Seebeck ing thermoelectric parts with any form or shape is a very effect. Nowadays, thermoelectric generators (TEG) are actual topic that has been differently addressed in other already powering a number of devices in a very broad research works, for instance, through the development of field of applications, ranging from medical, military and Bi Te -based inorganic paints with Sb Te as a sintering space applications, infrared sensors, computer chips, bat- 2 3 2 3 aid [10]. The Mg Si pastes herein proposed can be a com- tery charging, waste heat recovery (e.g. from car exhausts) petitive alternative applicable in a broad range of TEG- to rural home electrification [2 –4]. Although TEGs have based applications, e.g. from the automobile to the textile many advantages such as of compactness, low complex- sectors, here in the form of woven fabrics of functional ity, high reliability and silent operation (no moving parts), fibers. low maintenance cost and environmental compatibility of operation, they are not massively used due to their low TE conversion efficiency (< 10%). In fact, TEGs are actually Materials and methods used only in niche markets where the reliability is more important than performance and cost is not a main consid- Magnesium and silicon powders of less than 44 μm nominal eration [5]. Some of the issues with current TEGs hinder- grain size (mesh 325)—from Alfa Aesar with 99.8% and ing their proliferation are the lack of stability at extremes 99.5% purity, respectively—were loaded in a 2:1 mass ratio temperatures, along with problems of environmental into a 50 mL agate bowl along with hexane and three 20 mm friendliness, availability, and high costs of the base mate- diameter agate balls to be mechanically alloyed in a high rials and the synthesis. Therefore, materials such as M g Si energy planetary ball mill (Retsch PM100). Hexane was have recently attracted much attention: these alloys have added to prevent agglomeration of Mg powder on the walls been demonstrated as good TEG candidate base materials and milling balls. The fluid and balls-to-powder mass ratios as their synthesis has become easier and achievable by a were 2:1 and 10:1, respectively. To reduce and homogenize variety of methods, their constituent elements are non- the powder grain size, milling times of 2 h, 5 h and 10 h toxic (contrarily to direct competitors such as PbTe and were tested. The powders and hexane fluid were weighted CoSb ), abundant and light weight. The base silicide ther- and transferred to the mill bowl inside a glove box filled with moelectric properties can be enhanced and tuned through Ar gas. As the agate bowl is sealed inside the glove box with doping, increasing the conversion efficiency in many appli- an o-ring fitting lid secured by a custom-made clamp, Ar cations (e.g. industrial furnaces, automobile exhausts, and will also be the atmosphere inside the bowl during the mill, incinerators in the mid-temperature range 230–730°C). avoiding oxidation of the reactants. The rotational speed For instance, Mg Si doped with Sb, Al and Bi has been was 400 rpm with 5 min pauses every 30 min, in all cases; used for the low and high temperature ends, respectively after milling, the resulting powder was collected in the glove [6], while double doping allows higher figures of merit box and directly loaded in an alumina crucible for Ar ther- (ZT), currently in the range of 0.8–1.1 [7, 8], with new mal annealing at a flow rate of ~ 0.35 L/min. The holding developments promising ZT values higher than 1.6 [9]. temperature was set between 350 and 590°C, depending However, an important issue affecting Mg Si-based TEGs on the powders grain size. The annealing temperatures for is the lack of shape control of the traditional synthesis the unmilled (590 °C), and the 2 h, 5 h (410 °C) and 10 h methods, mostly relaying on ingots formation. This makes (350 °C) milled powders were defined from the differential it difficult or even impossible to properly adapt to curved scanning calorimetry (DSC) heat flow curves, simultane- heat sources, inevitably introducing higher thermal/heat ously performed with thermogravimetric measurements transfer impedances, leading to considerable heat losses (TGA). Both measurements were simultaneously performed 1 3 Materials for Renewable and Sustainable Energy (2019) 8:21 Page 3 of 10 21 in the thermal analyser STA 449 F3 Jupiter under differ - Gemini V-2380 surface area analyser from Micromeritics and ent atmospheres (air and N ) from room temperature up to Gemini v2.0 software. The specific surface area (Brunauer- 1000 °C, at a rate of 20 K/min. Emmett-Teller, BET, method) was determined from nitrogen For all samples, the annealing temperature profiles con- adsorption isotherms determination for samples immersed in sisted of a heating ramp of ~ 15 °C/min to the desired hold- a liquid nitrogen bath. Barrett-Joyner-Halenda (BJH) method ing temperature, and a holding time of ~ 75 min, after which was used to calculate the pore size distribution in the samples. the temperature was ramped down to ~ 160 °C. Then, the Prior to these measurements, the water vapor and adsorbed furnace was turned off and the powders left to cool to room gas were removed by purging the samples in nitrogen flow for temperature. about 10 h. Over this period, the heat treatment of samples The synthetized Mg Si powders were mixed with polysty- A and B was held at 120 °C, while for samples C at 300 °C. rene (PS)—from Sigma Aldrich, Mw ~ 350,000—in xylene The thermal conductivity was measured at room tempera- solutions for obtaining n-type thermoelectric paste formula- ture (300 K) using the Gustafsson Probe method (Hot Disk) tions—one per application method: 3D printing, spraying with the Thermal Constant Analyser TPS 2500 S. This method and electrospinning. For 3D printing, the Mg Si powders is based on the Transient Plane Source (TPS) technique and were blended with a solution of 20% wt of polystyrene in uses an electrically conductive double spiral flat sensor that is xylene in the mass proportions of 43/57 (formulation 1) and protected by a kapton 70 µm thick film, acting both as pulsed 40/60 (formulation 2). These formulations were extruded heat source and temperature sensor. The TPS was assembled in a home-adapted 3D printer equipped with a hot plate set between two similar 3D printed 10 mm diameter disks. The to 50 °C to favor the fast evaporation of xylene. Fibers of measured thermal conductivity is a result of 14 consecutive Mg Si were produced by low pressure N -spray gun (Wuto) and equal measurements. All measurement parameters were 2 2 using a diluted version of formulation 2 and by electrospin- double checked and the results were consistent, since the resid- ning using a blend of Mg Si powders with a 35% wt of PS uals of temperature data fitting as a function of time present solution in a mass proportion 7:93 (formulation 3). This a random scatter dispersion within a few 1.5 mK. This is also was loaded into a syringe (B. Braun) connected to a blunt indicative of a good contact between the sensor and the twin metallic needle with an internal diameter of 1.19 mm (18G samples, a stable temperature in the samples and that the heat from ITEC, Iberiana Technical). A syringe pump (NewEra pulse did not reach the sample boundary. SyringePump.com) was used to eject the solution at a con- Electric and thermoelectric characterizations were per- trollable speed (0.2 mL/h) through the needle while a high formed with a home-made setup illustrated in Fig.  1. A voltage of 20 kV was applied (Glassman high voltage–power temperature difference (ΔT) is imposed across the piece supply). A grounded Al static plate was placed at 15 cm thickness using a heat source of variable temperature (from from the needle to collect the fibers. A fourth paste was for - 130 to 230 °C in steps of 25 °C) and one TEC1-12707 Pel- mulated with polyvinylidene difluoride (PVDF) solution in tier module connected to an independent power source— dimethylformamide (DMF) and then tested to produce bulk meant to work as the cold source. The Seebeck coefficient Mg Si parts. The PVDF was heated together with DMF at was determined by imposing a ∆T and monitoring it with a 70 °C until PVDF is completely dissolved. FLIRA310 thermal camera, while the thermoelectric volt- The temperatures at which the PS polymer can be burned age (ΔV) was measured using an Agilent 34420A nano- out from the printed pieces were determined by DST/TGA voltmeter, using C-paste electrodes with conductive tapes to be ~ 460–470 °C (depending on the PS concentration). on top and connecting to the cold/hot sources. The electrode 2 2 A holding time of ~ 90 min preceded by a heating ramp of area was ~ 70 mm and was separated by ~ 9 mm . Seebeck 2–5 °C/min revealed to be enough for that end. coefficient was obtained from the slope of the plot ∆V versus The morphology and composition of the milled powders, ∆T as shown in Fig. 1, subsequently enabling the calculus of before and after annealing, were studied using a FEG-SEM the power factor. A variable load resistance connected to the Jeol JSM7001F and a Vega 3 TESCAN scanning electron TE elements and ΔV measurements across terminals, from microscopes (SEM), both equipped with an energy disper- short-circuit to open-circuit conditions enable power output sive X-ray spectrometer (EDS). The crystalline phases were determination (P = I × V ). out out out identified by X-ray diffraction (XRD) using a Panalytical X-PERT Powder diffraction unit, through Cu K radiation (λ = 0.1540598  nm). Confocal Raman spectrophotometer Results and discussion (Witec Alpha 300 RAS) using a laser with a wavelength of 532 nm and 4.11 mW of power was used to confirm the exist- Mg Si powder synthesis ence of the Mg Si phase on the synthetized powders and also on both the printed pieces and fibers. The surface area, the The SEM images shown in Fig. 2 illustrate the influence of pore volume and the average pore size were measured using ball milling in the particle size of Mg- and Si- powders. The 1 3 21 Page 4 of 10 Materials for Renewable and Sustainable Energy (2019) 8:21 purchased powders shown in figure (a) and (b) consist of irregularly shaped grains with a somewhat heterogeneous size distributions, with mean values and standard deviations of, respectively, ~ 13.5 μm and 7.5 μm for Si, and ~ 37.9 μm and 11.9  μm for Mg, in compliance with the stated 325 mesh specifications—implying a particle size distribution with upper limit of ~ 44 μm. After milling the Mg- and Si- powders in a 2:1 mass ratio for 2 h (Fig. 2c), 5 h (Fig. 2d) and 10 h (Fig. 2e), both shape and grain size distribution become more homogeneous, while the grain size was pro- gressively reduced to ~ 26.3 ± 24.9  μm, ~ 8.8 ± 5.3 μm and ~ 7 ± 6.7 μm, respectively. The milling of the Mg- and Si- powder mixture was performed not only for reducing and homogenizing the grain size, aiming at enhancing the extru- sion of pastes by 3D printing, spray and electrospinning, but also for verifying if any amount of the Mg Si phase had formed as a result of the relatively short duration of the ball milling. Usually, much longer durations are required [11, 12]. XRD measurements on pristine Mg and Si powders, and mixtures of both, followed by 2 h, 5 h and 10 h milling are shown in Fig. 2. In all diffratograms, only the diffrac- tion lines arising from Mg and Si are seen, no significant Fig. 1 A schematic of the cross-sectional view of the home-made additional phases were detected (MgO fraction was below apparatus for the Seebeck measurements of bulk thermoelectrics. 1, Nano-voltmeter probes to measure the thermo-voltage; 2, conductive to 3%). Regardless the milling time, the XRD diffratograms tapes attached to the C-paste-based electrodes on the cylinder faces; from the milled powders can be described as a linear com- 3, Mg Si thermoelectric cylinder; 4, cold peltier; 5, variable hot bination of XRD diffratograms of the pure Mg and Si com- source; 6, heat sink mercial powders, with no trace of Mg Si whatsoever. This observation, together with that presented in Fig. 2, clearly demonstrates that milling up to 10 h, does not lead to alloy- ing; it merely decreases the grain sizes and homogenizes Fig. 2 SEM micrographs of Si and Mg commercial powders before diffratograms of the starting Si and Mg powders (top) and the milled (a–b) and after milling for 2 h (c), 5 h (d) and 10 h (e). The average powders (bottom) are shown in (f) grain size is indicated together with the standard deviation. The XRD 1 3 Materials for Renewable and Sustainable Energy (2019) 8:21 Page 5 of 10 21 Fig. 3 In the left half: SEM micrographs of Mg Si powders obtained Mg Si powders (e) obtained directly by thermal annealing (pink line) 2 2 directly by thermal annealing only (a) and after the sequence ther- and by thermal annealing and subsequent 5 h milling (gray line). For mal annealing and 5  h milling (b); photos of extruded Mg Si paste comparison, the XRD measurements obtained from powders milled made with powders formed through thermal annealing only (c) and for 5 h and annealed is also included (blue line) close-up (d), respectively. In the right half: XRD diffratograms of the Table 1 XRD phases quantification of powders processed under dif- fraction of ~ 71.3% for this sequence. As shown in Table 1, ferent conditions similar fractions of Mg Si were also formed through thermal annealing only (~ 76.3%) or in the sequence Ar TA → 5 h Powder synthesis sequence: Qty. (%) BM (~ 78.3%). Additional phases are due to unreacted Mg Mg Si MgO Si Mg SiO 2 2 fast oxidization, leading to the MgO phase as well as unre- Ar TA 76.3 8.3 1.4 14.1 – acted Si, that can also react with O and lead to SiO . The 2 2 5 h BM + Ar TA 71.3 11.1 7.0 – 10.6 presence of some MgO along with the Mg Si is not entirely Ar TA + 5 h BM 78.3 6.8 4.8 10.1 – surprising, since the insertion of the powder carrying cruci- bles in the quartz tubes used for annealing was always done in open air. The sequence BM→ Ar TA was also performed for a the size distributions. On the contrary, Fig. 3 XRD diffrato- grams clearly show that thermal annealing is of paramount shorter milling duration of 2  h as shown in Fig.  4a, the XRD diffratograms are compared with that of Mg–Si pow - importance for the formation of the Mg Si phase. This can be easily obtained through thermal annealing (TA) only of ders processed in the sequence 5  h BM → Ar TA. Both diffratograms are very similar which suggests that milling the Mg- and Si- powders in a 2:1 mass ratio, even without ball milling (BM). The drawback is that Mg Si powders pro- duration only impacts, as expected, in the final grain sizes. However, thermal annealing further reduced down the grain duced in this way lead to a small amount of unreacted Mg and to grains agglomerations which, as illustrated in Fig. 3, sizes of Mg–Si powders produced through 2 h BM → Ar TA to 3 μm, and 5 h BM→Ar TA to 2 μm, respectively. This made it impossible to 3D print the pastes formulation with such powders on the available printer, currently operating is shown in the SEM micrographs of Fig. 4b, c and may be attributed to the release of high residual stresses which may with needles of inner diameter of 0.61 mm and 1.19 mm. For mitigating this issue, the Mg Si powders produced by have led to polygonization and hence to the formation of new small grains with more homogeneous microstructure. directly annealing the Si–Mg powder mix were subsequently milled for 5 h to reduce and homogenize the particles, mak- Mg Si paste synthesis and application ing them suitable for manufacturing 3D printing pastes. The XRD diffratogram of Mg Si powders produced in the reverse The high reactivity of Si and Mg powders with oxygen sequence (i.e. 5 h BM → Ar TA) is also included in Fig. 3 for comparison. This does not exhibit the peaks correspond- demands the use of solvents and polymers free of oxy- gen. Polystyrene ((C H )n) and polyvinylidene difluoride ing to a fraction of unreacted Mg and moreover, quantifica- 8 8 tion through the Rietveld ren fi ement method yields an Mg Si ((C H F )n-)) polymers proved to be viable polymers when 2 2 2 1 3 21 Page 6 of 10 Materials for Renewable and Sustainable Energy (2019) 8:21 Fig. 4 XRD diffratograms (a) of Si–Mg powders ball milled (BM) shown in b and c. The average grain size taken from each micrograph for 2  h (black line) and 5  h (blue line) and next thermally annealed is indicated together with the standard deviation in Ar. The SEM micrographs after the synthesis sequences are also mixed in the correct proportion with Xylene ( C H ) and Table 2 Mg Si paste formulations prepared with polystyrene (PS) 8 10 2 solution for 3D printing and fiber production by spray and electro- Dimethylformamide(C H NO), respectively. Although the 3 7 spinning latter has in its constitution oxygen, Mg Si is not sensitive to Paste ID Paste application PS solution: Mg Si PS% wt. O (only to moisture) and it is one of the most recommended 2 2 powder mass ratio in xylene solvents for effectively dissolving PVDF. Hence, using this polymer–solvent combination only requires assuring that the 1 3D printing 57:43 20 fraction of unreacted Mg and Si in the synthetized Mg Si 2 3D printing and 60:40 20 powder batch is inexistent or negligible. Also, PVDF should spray (diluted) not be discarded because of its difficulty in finding compati- 3 Electrospinning 93:7 20 ble solvents without oxygen, since it has attractive properties for the Mg Si pastes formulation. It is non-toxic, has good thermal stability up to 100 °C, melts at 170 °C, is resistant to powder and with 3% less amount of Mg Si powder (for- chemicals, may exist in different crystalline forms depending mulation 2). This latter formulation was next diluted for on the preparation conditions, and most importantly, it has producing Mg Si fibers by low N pressure spray with the 2 2 low water absorption characteristics [13]. Therefore, both Wuto gun and by electrospinning (formulation 3). The SEM polymeric solutions were mixed with the synthetized Mg Si images of the sprayed fibers are shown in Fig.  6b, c and evi- powders for obtaining n-type thermoelectric Mg Si paste dence that these are aligned and incorporate Mg Si grains. 2 2 formulations. Prior to the PVDF paste formulation, three Similarly, the SEM images of electrospun fibers shown in polystyrene paste formulations were derived and extruded by Fig. 6c, d reveal their elongated beaded-like morphology, 3D printing, spray and electrospinning. Each paste formula- that includes Mg Si aggregates. The Raman spectra meas- tion is described in Table 2, and the ability of producing 3D ured on the printed pieces and sprayed fibers obtained with printed pieces, with a variety of shapes and high finishing paste formulations 2 and 3, respectively, are also included −1 quality, and fibers is illustrated in Fig.  5. in Fig. 6. The Raman peaks at ~ 251 cm are assigned to For 3D printing, the paste was made of Mg Si powders Mg Si phonon peaks identified in the literature as arising at −1 obtained from different milling times (5 h and 2 h). Figure  6a 256–260 cm , being the shift probably due residual strain shows that the pieces produced with 5 h milled powder of [14, 15]. lower grain size (formulation 1) are less smooth and have a The porosity of three printed small cylindrical pieces worst finishing than those produced with higher grain size (made with formulation 2) was also determined by means 1 3 Materials for Renewable and Sustainable Energy (2019) 8:21 Page 7 of 10 21 Fig. 5 a Pieces printed with Mg Si pastes following formulations 1 and 2 and SEM images of sprayed (b, c) and electrospun PS/Mg Si fibers (d, 2 2 e) produced with paste formulations 2 and 3, respectively. 3D-printing formulations are specified in Table 2 expected, to a low thermal conductivity 0.226 ± 0.001 W/ mK, that is significantly lower than the experimental (range from 7.8 to 4.0W/mK at 323 K and 623 K, respectively, [16]) and theoretical (~ 9.5 to 10.5 W/mK at 300 K [17]) values reported to M g Si, but led to insulating and mechanically fragile pieces. For that reason, a new paste formulation was devised. This was constituted by a solution of PVDF in DMF high boiling point solvent (153 °C). The aim was to pre- vent larger pores formation due to rapid solvent evaporation before and during the annealing for the polymer removal. Besides, the amount of polymeric solution in the paste was Fig. 6 Thermo-voltage vs. temperature difference plot measured from an Mg Si pellet. From the linear fit slope, the Seebeck coefficient was reduced. The mass ratio of PVDF solution to Mg Si powder determined and enabled the calculus of the power factor, being both was optimized to 8/92 (formulation 4), which immediately values indicated led to the production of bulk Mg Si thermoelectric pieces with electrical resistance and impressive thermoelectric of BET measurements, preceded by a standard 10 h thermal properties. The polymeric solution is constituted by 6.6% wt of PVDF in DMF. treatment. Two pieces, A and B, were heated at 120 °C and one at 300 °C, piece C. BET measurements on pieces A and Thermoelectric characterization B yielded a mean specific surface area of 4.11 ± 0.67 m /g, a total pore specific volume of 0.0030 ± 0.0002 cm /g, and a The thermoelectric characterization of Mg Si powder and pore size of 6.00 ± 0.34 nm, while for piece C, these values proved to be significantly smaller: 14.9 ± 2.4 m /g, 0.0194 ± bulk pieces produced with the ‘PVDF in DMF’ formulation is next presented. 0.0013 cm /g and 7.20 ± 0.41 nm, respectively. The reason is mainly attributed to the polymer evaporation that accord- Mg Si pellet ing to the literature is foreseen to occur at 210–249 °C, and at ~ 470  °C as determined by DSC/TGA measurements. Approximately, 283  mg of Mg Si powder was 15 ton Temperature at which polymer removal is expected to be completed. Such porosity after the polymer removal, led as pressed to form a pellet with a diameter of ~ 12.84 mm and 1 3 21 Page 8 of 10 Materials for Renewable and Sustainable Energy (2019) 8:21 hence the lack of thermoelectric property stability before sintering. This piece and other new replicas were sintered through a pressureless sintering method that consisted in a long thermal annealing performed at 400 °C under inert atmosphere. However, this sintering method, which cannot be combined with mechanical pressure, proved to be ineffec- tive, without leading to the consolidation of the bulk Mg Si pieces produced with a mean grain size of 2–3 μm, even after two steps of 16 h. The porosity and strength proper- ties were not sufficiently enhanced since after each sinter - Fig. 7 Thermo-voltage vs. temperature plot measured from a bulk ing attempt, it was not possible to measure the electrical Mg Si piece made with PVDF in DMF solution resistance. The sample composition was not significantly altered for all sintering attempts performed at 400 °C up to a thickness of ~ 1.55 mm. Measurements of the pellet elec- 32 h. The Raman spectra in Fig. 8 show that the dominant −1 trical resistance and voltage under a temperature difference phase remains that of Mg Si (247 cm peak) which coex- −1 was plotted as shown in Fig. 6 for determining the Seebeck ists with very small amounts of MgO (1348 and 1577 cm coefficient, where the former one corresponds to 132 kΩ peaks). Above 400 °C, Mg Si is not thermally stable and and the later to 487 μV/K. This Seebeck coefficient value other compounds start to form ~ MgO at 465 °C, ~ SiO at is in line with those reported in the literature for sintered 710 °C and ~ Mg SiO at ~ 1000 °C. 2 4 Mg Si, circa 500 μV/K [18], assuring that the Mg Si pow- An alternative pressureless method that predictably may 2 2 der herein synthetized for formulating the pastes is thermo- be used is the hot isostatic pressing sintering method, com- electric. The measured Seebeck coefficient value is slightly patible with 3D shapes. This subjects a sample to both ele- smaller, because of the formation of other minor phases, that vated temperature and isostatic gas pressure in a heated high are: unreacted Si and Mg, and SiO and MgO—as previ- pressure vessel filled with an inert gas for avoiding chemical ously concluded, the powder is not pure Mg Si. The curve reactions. This synthesis method may be an alternative that of Fig. 6 not only enabled the calculus of the Seebeck coef- together with others may be worth to explore to overcome ficient, but also of the power factor, ~ 0.58 nW/mK. Mg Si pieces made with PVDF polymeric solution Figure 7 shows the thermo-voltage measured as a function of the temperature difference applied to a bulk cylindrical piece made of paste constituted by 92% of M g Si and 8% of solu- tion of PVDF in DMF prior sintering. This curve enabled the calculus of the Seebeck (~ 4866.40 μV/K) coefficient and of the power factor (~ 8.5 μW/mK) as illustrated in Fig. 7. The first is in the range of competitor materials such as BiTe and CoSb (~ 290–700 V/K) but contrary to CoSb (5 μW/ 3 3 mK), BiTe still presents a higher power factor (~ 0.25 mW/ mK) than that of the Mg Si bulk piece prior to the polymer removal. This is evidently contributing to the thermoelectric properties measured from such Mg Si bulk piece. A few days later, the polymer inside the piece must have degraded, because it was no longer possible to measure the electrical resistance from this sample. Porosity initially filled by the polymer must have been partially undone due to the poly- mer degradation, which prevented the electrical resistance measurement. According to the literature, during useful Fig. 8 Raman spectra measured from replicas Mg Si pieces made of life, polymers may be influenced by heat, oxygen, sunlight, 92% of Mg Si and 8% of solution of PVDF in DMF before and after sintering performed at 400  °C of different durations: 1  h, 16  h and mechanical stress, etc. Also solvents can be photo-oxidized, 32  h. Black and green curves correspond to one sample before and hydrolyzed or thermally decompose, and as a result degrade after sintering and pink curves correspond to another sample, first the polymer. Therefore, it is difficult to exactly determine sintered by 16 h and next to another 16 h sintering (the 32 h were not what may have caused the PVDF polymer degradation and followed) 1 3 Materials for Renewable and Sustainable Energy (2019) 8:21 Page 9 of 10 21 electrospinning apparatus at the Biomaterials Laboratory from Soft and this sintering challenge and become the silicide’s extrusion Bio-functional Materials Group (CENIMAT/I3 N). And A.C. Baptista viable in the production of bulk thermoelectric parts. also acknowledges FCT-MEC for her postdoctoral grant with reference SFRH/BPD/104407/2014. Conclusions Open Access This article is distributed under the terms of the Crea- tive Commons Attribution 4.0 International License (http://creat iveco mmons.or g/licenses/b y/4.0/), which permits unrestricted use, distribu- Magnesium silicide powders, Mg Si, for TE applications tion, and reproduction in any medium, provided you give appropriate were successfully synthesized by combining ball milling credit to the original author(s) and the source, provide a link to the and thermal annealing. Ball milling alone does not yield Creative Commons license, and indicate if changes were made. Mg Si, as evidenced by XRD analysis, but it is needed to homogenize the particle size distribution and bring its overall dimensions to values suitable for their use in pastes References compatible with techniques such as 3D printing, spray and 1. Roco, M.C., Mirkin, C.A., Hersam, M.C.: Nanotechnology electrospinning. Research Directions for Societal Needs in 2020: Retrospective Although we have proven that Mg Si may also be syn- and Outlook. Springer, Netherlands (2011) thetized directly by thermal annealing only (without mill- 2. Jeffrey Snyder, G., Toberer, E.S.: Complex thermoelectric mate- ing), this option requires higher processing temperature rials. Nature Mater. (2008). https ://doi.org/10.1038/nmat2 090 3. Li, P., Cai, L., Zhai, P., Tang, X., Zhang, Q., Niino, M.: Design and revealed to be not compatible with some 3D printers of a concentration solar thermoelectric generator. J. Electron. operating with needles of inner diameter ≤ 1.19 mm. Mater. 39, 1522–1530 (2010). https :// doi.or g /10.100 7/s116 6 The formulation of Mg Si pastes with polystyrene and 4-010-1279-0 xylene proved to be viable for producing thermoelectric 4. Vargas-Almeida, A., Olivares-Robles, M.A., Camacho-Medina, P.: Thermoelectric system in different thermal and electrical parts with varied shapes and fibers by means of 3D print- configurations: its impact in the figure of merit. Entropy 15, ing, low-pressure spray and electrospinning. Despite fea- 2162–2180 (2013). https ://doi.org/10.3390/e1506 2162 sible, the PS-based paste 3D printing gives rise to very 5. Riffat, S.B., Ma, X.: Thermoelectrics: a review of present and porous pieces, which hindered electrical and thermoelec- potential applications. Appl. Therm. Eng. 23, 913–935 (2003). https ://doi.org/10.1016/S1359 -4311(03)00012 -7 trically characterization. As a result, the Mg Si content in 6. Hayatsu, Y., Iida, T., Sakamoto, T., Kurosaki, S., Nishio, K., a new paste formulation was significantly increased and Kogo, Y., Takanashi, Y.: Fabrication of large sintered pellets the polymeric solution changed to PVDF in DMF solu- of Sb-doped n-type Mg Si using a plasma activated sintering tion. This proved to be a viable formulation to generate method. J. Solid State Chem. 193, 161–165 (2012). https://doi. org/10.1016/j.jssc.2012.07.008 bulk Mg Si pieces with good thermoelectric properties 7. Joshi, G., Lee, H., Lan, Y., Wang, X., Zhu, G., Wang, D., Gould, (i.e. large Seebeck coefficient of 4866 μV/K and power R.W., Cuff, D.C., Tang, M.Y., Dresselhaus, M.S., Chen, G., factor of 8.5 μW/mK). However, the performance suffers Ren, Z.: Enhanced thermoelectric figure-of-merit in nanostruc - from degradation over time, probably due to changes in tured p-type silicon germanium bulk alloys. Nano Lett. 8, 4670 (2008). https ://doi.org/10.1021/nl802 6795 the polymer properties. Also, the long pressureless sinter- 8. Poudel, B., Hao, Q., Ma, Y., Lan, Y., Minnich, A., Yu, B., Yan, ing, performed at 400 °C due to the low thermal stabil- X., Wang, D., Muto, A., Vashaee, D., Chen, X., Liu, J., Dressel- ity Mg Si, has not been successful, which demonstrated haus, M.S., Chen, G., Ren, Z.: High-thermoelectric performance that this sintering method does not allow consolidating of nanostructured bismuth antimony telluride bulk alloys. Sci- ence 320, 634 (2008). https://doi.or g/10.1126/science.11564 46 the 3D pieces through pores reduction. Definitely, other 9. Trivedi, S.B., Kutcher, S.W., Rosemeier, C.A., David, M., alternatives will have to be explored to enable the sili- Jogender, S.: Magnesium and manganese silicides for efficient cides’s pastes formulation to be used in multiple extrusion and low cost thermo-electric power generation. United States techniques such as 3D printing and fiber making, which 616, 21152 (2013). https ://doi.org/10.2172/11084 70 10. Park, S.H., Jo, S., Kwon, B., Kim, F., Ban, H.W., Lee, J.E., Gu, requires further optimization. D.H., Lee, S.H., Hwang, Y., Kim, J.-S., Hyun, D.-B., Lee, S., Choi, K.J., Jo, W., Son, J.S.: High-performance shape-engineer- Acknowledgements This work was mainly funded by H2020- able thermoelectric painting. Nature Commun. 7, 13403 (2016). ICT-2014-1, RIA, TransFlexTeg-645241, and ERC-CoG-2014, https ://doi.org/10.1038/ncomm s1340 3 CapTherPV, 647596, and partially funded by FEDER funds through 11. Loannou, M., Hatzikraniotis, E., Lioutas, Ch., Hassapis, Th, the COMPETE 2020 Program and National Funds through FCT— Altantzis, Th, Paraskevopoulos, K.M., Kyratsi, Th: Fabri- Portuguese Foundation for Science and Technology under the project cation of nanocrystalline Mg Si via ball milling: structural UID/CTM/50025/2013. And co-supported by: (1) FCT—Portugal, studies. Powder Technol. 272, 523–532 (2012). https ://doi. through the contracts UID/Multi/04349/2013 and POCI-01-0145- org/10.1016/j.powte c.2011.11.014 FEDER-016674 and (2) CICECO-Aveiro Institute of Materials 12. Wang, L., Qin, X.Y., Xiong, W., Chen, L., Kong, M.G.: Ther- through the project POCI-01-0145-FEDER-007679 (FCT ref. UID/ mal stability and grain growth behavior of nanocrystalline CTM/50011/2013), financed by national funds through the FCT/ Mg Si. Mater. Sci. Eng. A 434, 166–170 (2006). https ://doi. MEC and when appropriate co-financed by FEDER under the PT2020 org/10.1016/j.msea.2006.06.130 Partnership Agreement. The authors would like to thank the use of 1 3 21 Page 10 of 10 Materials for Renewable and Sustainable Energy (2019) 8:21 13. Thomas, P., Satapathy, S., Dwarakanath, K.D., Varma, K.B.R.: 17. Boulet, P., Verstraete, M.J., Crocombette, J.-P., Briki, M., Record, Dielectric properties of Ply(vinylidene fluoride)/CaCu Ti O M.-C.: Electronic properties of the Mg Si thermoelectric material 3 4 12 2 nanocrystal composite thick films. Express Polymer Lett. 2010, investigated by linear-response density-functional theory. Comput. 4 (2010). https ://doi.org/10.3144/expre sspol ymlet t.2010.78 Mater. Sci. 50, 847–851 (2011). https://doi.or g/10.1016/j.comma 14. Galkin, K.N., Galkin, N.G.: Silicon overgrowth atop low-dimen- tsci.2010.10.020 sional Mg Si on Si(111): structure, optical and thermoelectri- 18. Saleemi, M., Toprak, M.S., Fiameni, S., Boldrini, S., Battiston, S., cal properties. Phys. Procedia 11, 55–58 (2011). https ://doi. Famengo, A., Stingaciu, M., Johnsson, M., Muhammed, M.: Spark org/10.1016/j.phpro .2011.01.013 plasma sintering and thermoelectric evaluation of nanocrystalline 15. Dotsenko, S.A., Galkin, K.N., Bezbabny, D.A., Goroshko, D.L., magnesium silicide (Mg Si). J. Mater. Sci. 48, 1940–1946 (2013). Galkin, N.G.: Formation, optical and electrical properties of a https ://doi.org/10.1007/s1085 3-012-6959-0 new semiconductor phase of calcium silicide on Si(111). Phys. Procedia 23, 41–44 (2012). https ://doi.or g/10.1016/j.phpr o Publisher’s Note Springer Nature remains neutral with regard to .2012.01.011 jurisdictional claims in published maps and institutional affiliations. 16. Wunderlich, W., Suzuki, Y., Gibo, N., Ohkuma, T., Al-Abandi, M., Khan, A.U., Mori, T.: Thermoelectric properties of Mg Si produced by new chemical route and SPS. Inorganics 2, 351–362 (2014). https ://doi.org/10.3390/inorg anics 20203 51 1 3 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Materials for Renewable and Sustainable Energy Springer Journals

Synthesis of thermoelectric magnesium-silicide pastes for 3D printing, electrospinning and low-pressure spray

Loading next page...
 
/lp/springer-journals/synthesis-of-thermoelectric-magnesium-silicide-pastes-for-3d-printing-skAqz0OLI5

References (21)

Publisher
Springer Journals
Copyright
Copyright © 2019 by The Author(s)
Subject
Materials Science; Materials Science, general; Renewable and Green Energy
ISSN
2194-1459
eISSN
2194-1467
DOI
10.1007/s40243-019-0159-7
Publisher site
See Article on Publisher Site

Abstract

In this work, eco-friendly magnesium-silicide (Mg Si) semiconducting (n-type) thermoelectric pastes for building compo- nents concerning energy-harvesting devices through 3D printing, spray and electrospinning were synthetized and tested for the first time. The Mg Si fine powders were obtained through the combination of ball milling and thermal annealing under Ar atmosphere. While the latter process was crucial for obtaining the desired Mg Si phase, the ball milling was indispensable for homogenizing and reducing the grain size of the powders. The synthetized Mg Si powders exhibited a large Seebeck coef- ficient of ~ 487 µV/K and were blended with a polymeric solution in different mass ratios to adjust the paste viscosity to the different requirements of 3D printing, electrospinning and low-pressure spray. The materials produced in every single stage of the paste synthesis were characterized by a variety of techniques that unequivocally prove their viability for producing thermoelectric parts and components. These can certainly trigger further research and development in green thermoelectric generators (TEGs) capable of adopting any form or shape with enhanced thermoelectric properties. These green TEGs are meant to compete with common toxic materials such as Bi Te , PbTe and CoSb that have Seebeck coefficients in the range 2 3 of ~ 290–700 μV/K, similar to that of the produced Mg Si powders and lower than that of 3D printed bulk Mg Si pieces, 2 2 measured to be ~ 4866 μV/K. Also, their measured thermal conductivities proved to be significantly lower (~ 0.2 W/mK) than that reported for Mg Si (≥ 4 W/mK). However, it is herein demonstrated that such thermoelectric properties are not stable over time. Pressureless sintering proved to be indispensable, but dic ffi ultly achievable by long thermal annealing (even above 32 h) in inert atmosphere at 400 °C, at least for bulk Mg Si pieces constituted by a mean grain size of 2–3 μm. Hence, for overcoming this sintering challenge and become the silicide’s extrusion viable in the production of bulk thermoelectric parts, alternative pressureless sintering methods will have to be further explored. Keywords Mg Si-based thermoelectric materials · Mechanical alloying · 3D printing · Electrospinning · Spray * A. C. Marques Departamento de Física e CICECO, Instituto de Materiais de acl.marques@fct.unl.pt Aveiro, Universidade de Aveiro, 3810-193 Aveiro, Portugal C2TN, Instituto Superior Técnico, Universidade de Lisboa, CENIMAT/I3N, Departamento de Ciência dos Materiais, Campus Tecnológico e Nuclear, Estrada Nacional 10, Faculdade de Ciências e Tecnologia, Universidade Nova de 2695-066 Bobadela, LRS, Portugal Lisboa, 2829-516 Caparica, Portugal IPFN-IST/UL, Instituto de Plasmas e Fusão Nuclear, Instituto Dipartimento di Scienza Applicata e Tecnologia, Politecnico Superior Técnico, Universidade de Lisboa, Estrada Nacional di Torino, Corso Duca degli Abruzzi 24, 10129 Turin, Italy 10, 2695-066 Bobadela, Portugal I3N/Departamento de Física e CICECO, Instituto de Materiais de Aveiro, Universidade de Aveiro, 3810-193 Aveiro, Portugal Vol.:(0123456789) 1 3 21 Page 2 of 10 Materials for Renewable and Sustainable Energy (2019) 8:21 and lower energy conversion efficiencies of the devices. A Introduction new approach is herein devised to overcome the challenge of shape: it consists in the production of Mg Si powders The increasing energy demand worldwide has been driv- through a simple and cost-effective process (relying on the ing the search for new, clean, renewable and sustain- combination of ball milling with thermal annealing), for able energy sources. Solar, wind and hydropower energy subsequent formulation of thermoelectric pastes suitable sources are expected to fulfill future energy needs and for 3D printing, electrospinning and spray technologies. replace energy sources based on fossil fuels. However, The major problems with the Mg Si powder synthesis and currently, these still assure about 90% of the world’s elec- paste formulation are related to the high reactivity of Si tricity generation with low operating efficiency (30–40%) and Mg powders with oxygen, demanding the use of an and large annual waste of heat to the environment (15 inert atmosphere, e.g. a glove box filled with Ar, and lim- TW) [1]. Such large amount of wasted heat can be directly its the selection of solvents and polymers to oxygen-free converted into electricity by solid-state generators based compounds. One should note that the need for develop- on the thermoelectric (TE) materials, using the Seebeck ing thermoelectric parts with any form or shape is a very effect. Nowadays, thermoelectric generators (TEG) are actual topic that has been differently addressed in other already powering a number of devices in a very broad research works, for instance, through the development of field of applications, ranging from medical, military and Bi Te -based inorganic paints with Sb Te as a sintering space applications, infrared sensors, computer chips, bat- 2 3 2 3 aid [10]. The Mg Si pastes herein proposed can be a com- tery charging, waste heat recovery (e.g. from car exhausts) petitive alternative applicable in a broad range of TEG- to rural home electrification [2 –4]. Although TEGs have based applications, e.g. from the automobile to the textile many advantages such as of compactness, low complex- sectors, here in the form of woven fabrics of functional ity, high reliability and silent operation (no moving parts), fibers. low maintenance cost and environmental compatibility of operation, they are not massively used due to their low TE conversion efficiency (< 10%). In fact, TEGs are actually Materials and methods used only in niche markets where the reliability is more important than performance and cost is not a main consid- Magnesium and silicon powders of less than 44 μm nominal eration [5]. Some of the issues with current TEGs hinder- grain size (mesh 325)—from Alfa Aesar with 99.8% and ing their proliferation are the lack of stability at extremes 99.5% purity, respectively—were loaded in a 2:1 mass ratio temperatures, along with problems of environmental into a 50 mL agate bowl along with hexane and three 20 mm friendliness, availability, and high costs of the base mate- diameter agate balls to be mechanically alloyed in a high rials and the synthesis. Therefore, materials such as M g Si energy planetary ball mill (Retsch PM100). Hexane was have recently attracted much attention: these alloys have added to prevent agglomeration of Mg powder on the walls been demonstrated as good TEG candidate base materials and milling balls. The fluid and balls-to-powder mass ratios as their synthesis has become easier and achievable by a were 2:1 and 10:1, respectively. To reduce and homogenize variety of methods, their constituent elements are non- the powder grain size, milling times of 2 h, 5 h and 10 h toxic (contrarily to direct competitors such as PbTe and were tested. The powders and hexane fluid were weighted CoSb ), abundant and light weight. The base silicide ther- and transferred to the mill bowl inside a glove box filled with moelectric properties can be enhanced and tuned through Ar gas. As the agate bowl is sealed inside the glove box with doping, increasing the conversion efficiency in many appli- an o-ring fitting lid secured by a custom-made clamp, Ar cations (e.g. industrial furnaces, automobile exhausts, and will also be the atmosphere inside the bowl during the mill, incinerators in the mid-temperature range 230–730°C). avoiding oxidation of the reactants. The rotational speed For instance, Mg Si doped with Sb, Al and Bi has been was 400 rpm with 5 min pauses every 30 min, in all cases; used for the low and high temperature ends, respectively after milling, the resulting powder was collected in the glove [6], while double doping allows higher figures of merit box and directly loaded in an alumina crucible for Ar ther- (ZT), currently in the range of 0.8–1.1 [7, 8], with new mal annealing at a flow rate of ~ 0.35 L/min. The holding developments promising ZT values higher than 1.6 [9]. temperature was set between 350 and 590°C, depending However, an important issue affecting Mg Si-based TEGs on the powders grain size. The annealing temperatures for is the lack of shape control of the traditional synthesis the unmilled (590 °C), and the 2 h, 5 h (410 °C) and 10 h methods, mostly relaying on ingots formation. This makes (350 °C) milled powders were defined from the differential it difficult or even impossible to properly adapt to curved scanning calorimetry (DSC) heat flow curves, simultane- heat sources, inevitably introducing higher thermal/heat ously performed with thermogravimetric measurements transfer impedances, leading to considerable heat losses (TGA). Both measurements were simultaneously performed 1 3 Materials for Renewable and Sustainable Energy (2019) 8:21 Page 3 of 10 21 in the thermal analyser STA 449 F3 Jupiter under differ - Gemini V-2380 surface area analyser from Micromeritics and ent atmospheres (air and N ) from room temperature up to Gemini v2.0 software. The specific surface area (Brunauer- 1000 °C, at a rate of 20 K/min. Emmett-Teller, BET, method) was determined from nitrogen For all samples, the annealing temperature profiles con- adsorption isotherms determination for samples immersed in sisted of a heating ramp of ~ 15 °C/min to the desired hold- a liquid nitrogen bath. Barrett-Joyner-Halenda (BJH) method ing temperature, and a holding time of ~ 75 min, after which was used to calculate the pore size distribution in the samples. the temperature was ramped down to ~ 160 °C. Then, the Prior to these measurements, the water vapor and adsorbed furnace was turned off and the powders left to cool to room gas were removed by purging the samples in nitrogen flow for temperature. about 10 h. Over this period, the heat treatment of samples The synthetized Mg Si powders were mixed with polysty- A and B was held at 120 °C, while for samples C at 300 °C. rene (PS)—from Sigma Aldrich, Mw ~ 350,000—in xylene The thermal conductivity was measured at room tempera- solutions for obtaining n-type thermoelectric paste formula- ture (300 K) using the Gustafsson Probe method (Hot Disk) tions—one per application method: 3D printing, spraying with the Thermal Constant Analyser TPS 2500 S. This method and electrospinning. For 3D printing, the Mg Si powders is based on the Transient Plane Source (TPS) technique and were blended with a solution of 20% wt of polystyrene in uses an electrically conductive double spiral flat sensor that is xylene in the mass proportions of 43/57 (formulation 1) and protected by a kapton 70 µm thick film, acting both as pulsed 40/60 (formulation 2). These formulations were extruded heat source and temperature sensor. The TPS was assembled in a home-adapted 3D printer equipped with a hot plate set between two similar 3D printed 10 mm diameter disks. The to 50 °C to favor the fast evaporation of xylene. Fibers of measured thermal conductivity is a result of 14 consecutive Mg Si were produced by low pressure N -spray gun (Wuto) and equal measurements. All measurement parameters were 2 2 using a diluted version of formulation 2 and by electrospin- double checked and the results were consistent, since the resid- ning using a blend of Mg Si powders with a 35% wt of PS uals of temperature data fitting as a function of time present solution in a mass proportion 7:93 (formulation 3). This a random scatter dispersion within a few 1.5 mK. This is also was loaded into a syringe (B. Braun) connected to a blunt indicative of a good contact between the sensor and the twin metallic needle with an internal diameter of 1.19 mm (18G samples, a stable temperature in the samples and that the heat from ITEC, Iberiana Technical). A syringe pump (NewEra pulse did not reach the sample boundary. SyringePump.com) was used to eject the solution at a con- Electric and thermoelectric characterizations were per- trollable speed (0.2 mL/h) through the needle while a high formed with a home-made setup illustrated in Fig.  1. A voltage of 20 kV was applied (Glassman high voltage–power temperature difference (ΔT) is imposed across the piece supply). A grounded Al static plate was placed at 15 cm thickness using a heat source of variable temperature (from from the needle to collect the fibers. A fourth paste was for - 130 to 230 °C in steps of 25 °C) and one TEC1-12707 Pel- mulated with polyvinylidene difluoride (PVDF) solution in tier module connected to an independent power source— dimethylformamide (DMF) and then tested to produce bulk meant to work as the cold source. The Seebeck coefficient Mg Si parts. The PVDF was heated together with DMF at was determined by imposing a ∆T and monitoring it with a 70 °C until PVDF is completely dissolved. FLIRA310 thermal camera, while the thermoelectric volt- The temperatures at which the PS polymer can be burned age (ΔV) was measured using an Agilent 34420A nano- out from the printed pieces were determined by DST/TGA voltmeter, using C-paste electrodes with conductive tapes to be ~ 460–470 °C (depending on the PS concentration). on top and connecting to the cold/hot sources. The electrode 2 2 A holding time of ~ 90 min preceded by a heating ramp of area was ~ 70 mm and was separated by ~ 9 mm . Seebeck 2–5 °C/min revealed to be enough for that end. coefficient was obtained from the slope of the plot ∆V versus The morphology and composition of the milled powders, ∆T as shown in Fig. 1, subsequently enabling the calculus of before and after annealing, were studied using a FEG-SEM the power factor. A variable load resistance connected to the Jeol JSM7001F and a Vega 3 TESCAN scanning electron TE elements and ΔV measurements across terminals, from microscopes (SEM), both equipped with an energy disper- short-circuit to open-circuit conditions enable power output sive X-ray spectrometer (EDS). The crystalline phases were determination (P = I × V ). out out out identified by X-ray diffraction (XRD) using a Panalytical X-PERT Powder diffraction unit, through Cu K radiation (λ = 0.1540598  nm). Confocal Raman spectrophotometer Results and discussion (Witec Alpha 300 RAS) using a laser with a wavelength of 532 nm and 4.11 mW of power was used to confirm the exist- Mg Si powder synthesis ence of the Mg Si phase on the synthetized powders and also on both the printed pieces and fibers. The surface area, the The SEM images shown in Fig. 2 illustrate the influence of pore volume and the average pore size were measured using ball milling in the particle size of Mg- and Si- powders. The 1 3 21 Page 4 of 10 Materials for Renewable and Sustainable Energy (2019) 8:21 purchased powders shown in figure (a) and (b) consist of irregularly shaped grains with a somewhat heterogeneous size distributions, with mean values and standard deviations of, respectively, ~ 13.5 μm and 7.5 μm for Si, and ~ 37.9 μm and 11.9  μm for Mg, in compliance with the stated 325 mesh specifications—implying a particle size distribution with upper limit of ~ 44 μm. After milling the Mg- and Si- powders in a 2:1 mass ratio for 2 h (Fig. 2c), 5 h (Fig. 2d) and 10 h (Fig. 2e), both shape and grain size distribution become more homogeneous, while the grain size was pro- gressively reduced to ~ 26.3 ± 24.9  μm, ~ 8.8 ± 5.3 μm and ~ 7 ± 6.7 μm, respectively. The milling of the Mg- and Si- powder mixture was performed not only for reducing and homogenizing the grain size, aiming at enhancing the extru- sion of pastes by 3D printing, spray and electrospinning, but also for verifying if any amount of the Mg Si phase had formed as a result of the relatively short duration of the ball milling. Usually, much longer durations are required [11, 12]. XRD measurements on pristine Mg and Si powders, and mixtures of both, followed by 2 h, 5 h and 10 h milling are shown in Fig. 2. In all diffratograms, only the diffrac- tion lines arising from Mg and Si are seen, no significant Fig. 1 A schematic of the cross-sectional view of the home-made additional phases were detected (MgO fraction was below apparatus for the Seebeck measurements of bulk thermoelectrics. 1, Nano-voltmeter probes to measure the thermo-voltage; 2, conductive to 3%). Regardless the milling time, the XRD diffratograms tapes attached to the C-paste-based electrodes on the cylinder faces; from the milled powders can be described as a linear com- 3, Mg Si thermoelectric cylinder; 4, cold peltier; 5, variable hot bination of XRD diffratograms of the pure Mg and Si com- source; 6, heat sink mercial powders, with no trace of Mg Si whatsoever. This observation, together with that presented in Fig. 2, clearly demonstrates that milling up to 10 h, does not lead to alloy- ing; it merely decreases the grain sizes and homogenizes Fig. 2 SEM micrographs of Si and Mg commercial powders before diffratograms of the starting Si and Mg powders (top) and the milled (a–b) and after milling for 2 h (c), 5 h (d) and 10 h (e). The average powders (bottom) are shown in (f) grain size is indicated together with the standard deviation. The XRD 1 3 Materials for Renewable and Sustainable Energy (2019) 8:21 Page 5 of 10 21 Fig. 3 In the left half: SEM micrographs of Mg Si powders obtained Mg Si powders (e) obtained directly by thermal annealing (pink line) 2 2 directly by thermal annealing only (a) and after the sequence ther- and by thermal annealing and subsequent 5 h milling (gray line). For mal annealing and 5  h milling (b); photos of extruded Mg Si paste comparison, the XRD measurements obtained from powders milled made with powders formed through thermal annealing only (c) and for 5 h and annealed is also included (blue line) close-up (d), respectively. In the right half: XRD diffratograms of the Table 1 XRD phases quantification of powders processed under dif- fraction of ~ 71.3% for this sequence. As shown in Table 1, ferent conditions similar fractions of Mg Si were also formed through thermal annealing only (~ 76.3%) or in the sequence Ar TA → 5 h Powder synthesis sequence: Qty. (%) BM (~ 78.3%). Additional phases are due to unreacted Mg Mg Si MgO Si Mg SiO 2 2 fast oxidization, leading to the MgO phase as well as unre- Ar TA 76.3 8.3 1.4 14.1 – acted Si, that can also react with O and lead to SiO . The 2 2 5 h BM + Ar TA 71.3 11.1 7.0 – 10.6 presence of some MgO along with the Mg Si is not entirely Ar TA + 5 h BM 78.3 6.8 4.8 10.1 – surprising, since the insertion of the powder carrying cruci- bles in the quartz tubes used for annealing was always done in open air. The sequence BM→ Ar TA was also performed for a the size distributions. On the contrary, Fig. 3 XRD diffrato- grams clearly show that thermal annealing is of paramount shorter milling duration of 2  h as shown in Fig.  4a, the XRD diffratograms are compared with that of Mg–Si pow - importance for the formation of the Mg Si phase. This can be easily obtained through thermal annealing (TA) only of ders processed in the sequence 5  h BM → Ar TA. Both diffratograms are very similar which suggests that milling the Mg- and Si- powders in a 2:1 mass ratio, even without ball milling (BM). The drawback is that Mg Si powders pro- duration only impacts, as expected, in the final grain sizes. However, thermal annealing further reduced down the grain duced in this way lead to a small amount of unreacted Mg and to grains agglomerations which, as illustrated in Fig. 3, sizes of Mg–Si powders produced through 2 h BM → Ar TA to 3 μm, and 5 h BM→Ar TA to 2 μm, respectively. This made it impossible to 3D print the pastes formulation with such powders on the available printer, currently operating is shown in the SEM micrographs of Fig. 4b, c and may be attributed to the release of high residual stresses which may with needles of inner diameter of 0.61 mm and 1.19 mm. For mitigating this issue, the Mg Si powders produced by have led to polygonization and hence to the formation of new small grains with more homogeneous microstructure. directly annealing the Si–Mg powder mix were subsequently milled for 5 h to reduce and homogenize the particles, mak- Mg Si paste synthesis and application ing them suitable for manufacturing 3D printing pastes. The XRD diffratogram of Mg Si powders produced in the reverse The high reactivity of Si and Mg powders with oxygen sequence (i.e. 5 h BM → Ar TA) is also included in Fig. 3 for comparison. This does not exhibit the peaks correspond- demands the use of solvents and polymers free of oxy- gen. Polystyrene ((C H )n) and polyvinylidene difluoride ing to a fraction of unreacted Mg and moreover, quantifica- 8 8 tion through the Rietveld ren fi ement method yields an Mg Si ((C H F )n-)) polymers proved to be viable polymers when 2 2 2 1 3 21 Page 6 of 10 Materials for Renewable and Sustainable Energy (2019) 8:21 Fig. 4 XRD diffratograms (a) of Si–Mg powders ball milled (BM) shown in b and c. The average grain size taken from each micrograph for 2  h (black line) and 5  h (blue line) and next thermally annealed is indicated together with the standard deviation in Ar. The SEM micrographs after the synthesis sequences are also mixed in the correct proportion with Xylene ( C H ) and Table 2 Mg Si paste formulations prepared with polystyrene (PS) 8 10 2 solution for 3D printing and fiber production by spray and electro- Dimethylformamide(C H NO), respectively. Although the 3 7 spinning latter has in its constitution oxygen, Mg Si is not sensitive to Paste ID Paste application PS solution: Mg Si PS% wt. O (only to moisture) and it is one of the most recommended 2 2 powder mass ratio in xylene solvents for effectively dissolving PVDF. Hence, using this polymer–solvent combination only requires assuring that the 1 3D printing 57:43 20 fraction of unreacted Mg and Si in the synthetized Mg Si 2 3D printing and 60:40 20 powder batch is inexistent or negligible. Also, PVDF should spray (diluted) not be discarded because of its difficulty in finding compati- 3 Electrospinning 93:7 20 ble solvents without oxygen, since it has attractive properties for the Mg Si pastes formulation. It is non-toxic, has good thermal stability up to 100 °C, melts at 170 °C, is resistant to powder and with 3% less amount of Mg Si powder (for- chemicals, may exist in different crystalline forms depending mulation 2). This latter formulation was next diluted for on the preparation conditions, and most importantly, it has producing Mg Si fibers by low N pressure spray with the 2 2 low water absorption characteristics [13]. Therefore, both Wuto gun and by electrospinning (formulation 3). The SEM polymeric solutions were mixed with the synthetized Mg Si images of the sprayed fibers are shown in Fig.  6b, c and evi- powders for obtaining n-type thermoelectric Mg Si paste dence that these are aligned and incorporate Mg Si grains. 2 2 formulations. Prior to the PVDF paste formulation, three Similarly, the SEM images of electrospun fibers shown in polystyrene paste formulations were derived and extruded by Fig. 6c, d reveal their elongated beaded-like morphology, 3D printing, spray and electrospinning. Each paste formula- that includes Mg Si aggregates. The Raman spectra meas- tion is described in Table 2, and the ability of producing 3D ured on the printed pieces and sprayed fibers obtained with printed pieces, with a variety of shapes and high finishing paste formulations 2 and 3, respectively, are also included −1 quality, and fibers is illustrated in Fig.  5. in Fig. 6. The Raman peaks at ~ 251 cm are assigned to For 3D printing, the paste was made of Mg Si powders Mg Si phonon peaks identified in the literature as arising at −1 obtained from different milling times (5 h and 2 h). Figure  6a 256–260 cm , being the shift probably due residual strain shows that the pieces produced with 5 h milled powder of [14, 15]. lower grain size (formulation 1) are less smooth and have a The porosity of three printed small cylindrical pieces worst finishing than those produced with higher grain size (made with formulation 2) was also determined by means 1 3 Materials for Renewable and Sustainable Energy (2019) 8:21 Page 7 of 10 21 Fig. 5 a Pieces printed with Mg Si pastes following formulations 1 and 2 and SEM images of sprayed (b, c) and electrospun PS/Mg Si fibers (d, 2 2 e) produced with paste formulations 2 and 3, respectively. 3D-printing formulations are specified in Table 2 expected, to a low thermal conductivity 0.226 ± 0.001 W/ mK, that is significantly lower than the experimental (range from 7.8 to 4.0W/mK at 323 K and 623 K, respectively, [16]) and theoretical (~ 9.5 to 10.5 W/mK at 300 K [17]) values reported to M g Si, but led to insulating and mechanically fragile pieces. For that reason, a new paste formulation was devised. This was constituted by a solution of PVDF in DMF high boiling point solvent (153 °C). The aim was to pre- vent larger pores formation due to rapid solvent evaporation before and during the annealing for the polymer removal. Besides, the amount of polymeric solution in the paste was Fig. 6 Thermo-voltage vs. temperature difference plot measured from an Mg Si pellet. From the linear fit slope, the Seebeck coefficient was reduced. The mass ratio of PVDF solution to Mg Si powder determined and enabled the calculus of the power factor, being both was optimized to 8/92 (formulation 4), which immediately values indicated led to the production of bulk Mg Si thermoelectric pieces with electrical resistance and impressive thermoelectric of BET measurements, preceded by a standard 10 h thermal properties. The polymeric solution is constituted by 6.6% wt of PVDF in DMF. treatment. Two pieces, A and B, were heated at 120 °C and one at 300 °C, piece C. BET measurements on pieces A and Thermoelectric characterization B yielded a mean specific surface area of 4.11 ± 0.67 m /g, a total pore specific volume of 0.0030 ± 0.0002 cm /g, and a The thermoelectric characterization of Mg Si powder and pore size of 6.00 ± 0.34 nm, while for piece C, these values proved to be significantly smaller: 14.9 ± 2.4 m /g, 0.0194 ± bulk pieces produced with the ‘PVDF in DMF’ formulation is next presented. 0.0013 cm /g and 7.20 ± 0.41 nm, respectively. The reason is mainly attributed to the polymer evaporation that accord- Mg Si pellet ing to the literature is foreseen to occur at 210–249 °C, and at ~ 470  °C as determined by DSC/TGA measurements. Approximately, 283  mg of Mg Si powder was 15 ton Temperature at which polymer removal is expected to be completed. Such porosity after the polymer removal, led as pressed to form a pellet with a diameter of ~ 12.84 mm and 1 3 21 Page 8 of 10 Materials for Renewable and Sustainable Energy (2019) 8:21 hence the lack of thermoelectric property stability before sintering. This piece and other new replicas were sintered through a pressureless sintering method that consisted in a long thermal annealing performed at 400 °C under inert atmosphere. However, this sintering method, which cannot be combined with mechanical pressure, proved to be ineffec- tive, without leading to the consolidation of the bulk Mg Si pieces produced with a mean grain size of 2–3 μm, even after two steps of 16 h. The porosity and strength proper- ties were not sufficiently enhanced since after each sinter - Fig. 7 Thermo-voltage vs. temperature plot measured from a bulk ing attempt, it was not possible to measure the electrical Mg Si piece made with PVDF in DMF solution resistance. The sample composition was not significantly altered for all sintering attempts performed at 400 °C up to a thickness of ~ 1.55 mm. Measurements of the pellet elec- 32 h. The Raman spectra in Fig. 8 show that the dominant −1 trical resistance and voltage under a temperature difference phase remains that of Mg Si (247 cm peak) which coex- −1 was plotted as shown in Fig. 6 for determining the Seebeck ists with very small amounts of MgO (1348 and 1577 cm coefficient, where the former one corresponds to 132 kΩ peaks). Above 400 °C, Mg Si is not thermally stable and and the later to 487 μV/K. This Seebeck coefficient value other compounds start to form ~ MgO at 465 °C, ~ SiO at is in line with those reported in the literature for sintered 710 °C and ~ Mg SiO at ~ 1000 °C. 2 4 Mg Si, circa 500 μV/K [18], assuring that the Mg Si pow- An alternative pressureless method that predictably may 2 2 der herein synthetized for formulating the pastes is thermo- be used is the hot isostatic pressing sintering method, com- electric. The measured Seebeck coefficient value is slightly patible with 3D shapes. This subjects a sample to both ele- smaller, because of the formation of other minor phases, that vated temperature and isostatic gas pressure in a heated high are: unreacted Si and Mg, and SiO and MgO—as previ- pressure vessel filled with an inert gas for avoiding chemical ously concluded, the powder is not pure Mg Si. The curve reactions. This synthesis method may be an alternative that of Fig. 6 not only enabled the calculus of the Seebeck coef- together with others may be worth to explore to overcome ficient, but also of the power factor, ~ 0.58 nW/mK. Mg Si pieces made with PVDF polymeric solution Figure 7 shows the thermo-voltage measured as a function of the temperature difference applied to a bulk cylindrical piece made of paste constituted by 92% of M g Si and 8% of solu- tion of PVDF in DMF prior sintering. This curve enabled the calculus of the Seebeck (~ 4866.40 μV/K) coefficient and of the power factor (~ 8.5 μW/mK) as illustrated in Fig. 7. The first is in the range of competitor materials such as BiTe and CoSb (~ 290–700 V/K) but contrary to CoSb (5 μW/ 3 3 mK), BiTe still presents a higher power factor (~ 0.25 mW/ mK) than that of the Mg Si bulk piece prior to the polymer removal. This is evidently contributing to the thermoelectric properties measured from such Mg Si bulk piece. A few days later, the polymer inside the piece must have degraded, because it was no longer possible to measure the electrical resistance from this sample. Porosity initially filled by the polymer must have been partially undone due to the poly- mer degradation, which prevented the electrical resistance measurement. According to the literature, during useful Fig. 8 Raman spectra measured from replicas Mg Si pieces made of life, polymers may be influenced by heat, oxygen, sunlight, 92% of Mg Si and 8% of solution of PVDF in DMF before and after sintering performed at 400  °C of different durations: 1  h, 16  h and mechanical stress, etc. Also solvents can be photo-oxidized, 32  h. Black and green curves correspond to one sample before and hydrolyzed or thermally decompose, and as a result degrade after sintering and pink curves correspond to another sample, first the polymer. Therefore, it is difficult to exactly determine sintered by 16 h and next to another 16 h sintering (the 32 h were not what may have caused the PVDF polymer degradation and followed) 1 3 Materials for Renewable and Sustainable Energy (2019) 8:21 Page 9 of 10 21 electrospinning apparatus at the Biomaterials Laboratory from Soft and this sintering challenge and become the silicide’s extrusion Bio-functional Materials Group (CENIMAT/I3 N). And A.C. Baptista viable in the production of bulk thermoelectric parts. also acknowledges FCT-MEC for her postdoctoral grant with reference SFRH/BPD/104407/2014. Conclusions Open Access This article is distributed under the terms of the Crea- tive Commons Attribution 4.0 International License (http://creat iveco mmons.or g/licenses/b y/4.0/), which permits unrestricted use, distribu- Magnesium silicide powders, Mg Si, for TE applications tion, and reproduction in any medium, provided you give appropriate were successfully synthesized by combining ball milling credit to the original author(s) and the source, provide a link to the and thermal annealing. Ball milling alone does not yield Creative Commons license, and indicate if changes were made. Mg Si, as evidenced by XRD analysis, but it is needed to homogenize the particle size distribution and bring its overall dimensions to values suitable for their use in pastes References compatible with techniques such as 3D printing, spray and 1. Roco, M.C., Mirkin, C.A., Hersam, M.C.: Nanotechnology electrospinning. Research Directions for Societal Needs in 2020: Retrospective Although we have proven that Mg Si may also be syn- and Outlook. Springer, Netherlands (2011) thetized directly by thermal annealing only (without mill- 2. Jeffrey Snyder, G., Toberer, E.S.: Complex thermoelectric mate- ing), this option requires higher processing temperature rials. Nature Mater. (2008). https ://doi.org/10.1038/nmat2 090 3. Li, P., Cai, L., Zhai, P., Tang, X., Zhang, Q., Niino, M.: Design and revealed to be not compatible with some 3D printers of a concentration solar thermoelectric generator. J. Electron. operating with needles of inner diameter ≤ 1.19 mm. Mater. 39, 1522–1530 (2010). https :// doi.or g /10.100 7/s116 6 The formulation of Mg Si pastes with polystyrene and 4-010-1279-0 xylene proved to be viable for producing thermoelectric 4. Vargas-Almeida, A., Olivares-Robles, M.A., Camacho-Medina, P.: Thermoelectric system in different thermal and electrical parts with varied shapes and fibers by means of 3D print- configurations: its impact in the figure of merit. Entropy 15, ing, low-pressure spray and electrospinning. Despite fea- 2162–2180 (2013). https ://doi.org/10.3390/e1506 2162 sible, the PS-based paste 3D printing gives rise to very 5. Riffat, S.B., Ma, X.: Thermoelectrics: a review of present and porous pieces, which hindered electrical and thermoelec- potential applications. Appl. Therm. Eng. 23, 913–935 (2003). https ://doi.org/10.1016/S1359 -4311(03)00012 -7 trically characterization. As a result, the Mg Si content in 6. Hayatsu, Y., Iida, T., Sakamoto, T., Kurosaki, S., Nishio, K., a new paste formulation was significantly increased and Kogo, Y., Takanashi, Y.: Fabrication of large sintered pellets the polymeric solution changed to PVDF in DMF solu- of Sb-doped n-type Mg Si using a plasma activated sintering tion. This proved to be a viable formulation to generate method. J. Solid State Chem. 193, 161–165 (2012). https://doi. org/10.1016/j.jssc.2012.07.008 bulk Mg Si pieces with good thermoelectric properties 7. Joshi, G., Lee, H., Lan, Y., Wang, X., Zhu, G., Wang, D., Gould, (i.e. large Seebeck coefficient of 4866 μV/K and power R.W., Cuff, D.C., Tang, M.Y., Dresselhaus, M.S., Chen, G., factor of 8.5 μW/mK). However, the performance suffers Ren, Z.: Enhanced thermoelectric figure-of-merit in nanostruc - from degradation over time, probably due to changes in tured p-type silicon germanium bulk alloys. Nano Lett. 8, 4670 (2008). https ://doi.org/10.1021/nl802 6795 the polymer properties. Also, the long pressureless sinter- 8. Poudel, B., Hao, Q., Ma, Y., Lan, Y., Minnich, A., Yu, B., Yan, ing, performed at 400 °C due to the low thermal stabil- X., Wang, D., Muto, A., Vashaee, D., Chen, X., Liu, J., Dressel- ity Mg Si, has not been successful, which demonstrated haus, M.S., Chen, G., Ren, Z.: High-thermoelectric performance that this sintering method does not allow consolidating of nanostructured bismuth antimony telluride bulk alloys. Sci- ence 320, 634 (2008). https://doi.or g/10.1126/science.11564 46 the 3D pieces through pores reduction. Definitely, other 9. Trivedi, S.B., Kutcher, S.W., Rosemeier, C.A., David, M., alternatives will have to be explored to enable the sili- Jogender, S.: Magnesium and manganese silicides for efficient cides’s pastes formulation to be used in multiple extrusion and low cost thermo-electric power generation. United States techniques such as 3D printing and fiber making, which 616, 21152 (2013). https ://doi.org/10.2172/11084 70 10. Park, S.H., Jo, S., Kwon, B., Kim, F., Ban, H.W., Lee, J.E., Gu, requires further optimization. D.H., Lee, S.H., Hwang, Y., Kim, J.-S., Hyun, D.-B., Lee, S., Choi, K.J., Jo, W., Son, J.S.: High-performance shape-engineer- Acknowledgements This work was mainly funded by H2020- able thermoelectric painting. Nature Commun. 7, 13403 (2016). ICT-2014-1, RIA, TransFlexTeg-645241, and ERC-CoG-2014, https ://doi.org/10.1038/ncomm s1340 3 CapTherPV, 647596, and partially funded by FEDER funds through 11. Loannou, M., Hatzikraniotis, E., Lioutas, Ch., Hassapis, Th, the COMPETE 2020 Program and National Funds through FCT— Altantzis, Th, Paraskevopoulos, K.M., Kyratsi, Th: Fabri- Portuguese Foundation for Science and Technology under the project cation of nanocrystalline Mg Si via ball milling: structural UID/CTM/50025/2013. And co-supported by: (1) FCT—Portugal, studies. Powder Technol. 272, 523–532 (2012). https ://doi. through the contracts UID/Multi/04349/2013 and POCI-01-0145- org/10.1016/j.powte c.2011.11.014 FEDER-016674 and (2) CICECO-Aveiro Institute of Materials 12. Wang, L., Qin, X.Y., Xiong, W., Chen, L., Kong, M.G.: Ther- through the project POCI-01-0145-FEDER-007679 (FCT ref. UID/ mal stability and grain growth behavior of nanocrystalline CTM/50011/2013), financed by national funds through the FCT/ Mg Si. Mater. Sci. Eng. A 434, 166–170 (2006). https ://doi. MEC and when appropriate co-financed by FEDER under the PT2020 org/10.1016/j.msea.2006.06.130 Partnership Agreement. The authors would like to thank the use of 1 3 21 Page 10 of 10 Materials for Renewable and Sustainable Energy (2019) 8:21 13. Thomas, P., Satapathy, S., Dwarakanath, K.D., Varma, K.B.R.: 17. Boulet, P., Verstraete, M.J., Crocombette, J.-P., Briki, M., Record, Dielectric properties of Ply(vinylidene fluoride)/CaCu Ti O M.-C.: Electronic properties of the Mg Si thermoelectric material 3 4 12 2 nanocrystal composite thick films. Express Polymer Lett. 2010, investigated by linear-response density-functional theory. Comput. 4 (2010). https ://doi.org/10.3144/expre sspol ymlet t.2010.78 Mater. Sci. 50, 847–851 (2011). https://doi.or g/10.1016/j.comma 14. Galkin, K.N., Galkin, N.G.: Silicon overgrowth atop low-dimen- tsci.2010.10.020 sional Mg Si on Si(111): structure, optical and thermoelectri- 18. Saleemi, M., Toprak, M.S., Fiameni, S., Boldrini, S., Battiston, S., cal properties. Phys. Procedia 11, 55–58 (2011). https ://doi. Famengo, A., Stingaciu, M., Johnsson, M., Muhammed, M.: Spark org/10.1016/j.phpro .2011.01.013 plasma sintering and thermoelectric evaluation of nanocrystalline 15. Dotsenko, S.A., Galkin, K.N., Bezbabny, D.A., Goroshko, D.L., magnesium silicide (Mg Si). J. Mater. Sci. 48, 1940–1946 (2013). Galkin, N.G.: Formation, optical and electrical properties of a https ://doi.org/10.1007/s1085 3-012-6959-0 new semiconductor phase of calcium silicide on Si(111). Phys. Procedia 23, 41–44 (2012). https ://doi.or g/10.1016/j.phpr o Publisher’s Note Springer Nature remains neutral with regard to .2012.01.011 jurisdictional claims in published maps and institutional affiliations. 16. Wunderlich, W., Suzuki, Y., Gibo, N., Ohkuma, T., Al-Abandi, M., Khan, A.U., Mori, T.: Thermoelectric properties of Mg Si produced by new chemical route and SPS. Inorganics 2, 351–362 (2014). https ://doi.org/10.3390/inorg anics 20203 51 1 3

Journal

Materials for Renewable and Sustainable EnergySpringer Journals

Published: Oct 9, 2019

There are no references for this article.