Skip to main content

Nanostructured energy-harvesting thermoelectrics based on Mg2Si

Final Report Summary - THERMOMAG (Nanostructured energy-harvesting thermoelectrics based on Mg2Si)

Executive Summary:
The development of efficient thermoelectric devices for scavenging waste heat using low-cost, readily available, renewable and sustainable materials would certainly contribute to address a range of major challenges impacting on energy efficiency, climate change and resources depletion. Within this context, the core concept of the ThermoMag project revolved around developing and delivering sustainable and scalable energy harvesting thermoelectric (TE) materials and modules, based on nanostructured bulk Mg2Si solid solutions. The main scientific objective of this 3.5 years project has been to reveal a profound and quantified understanding of the links between the starting composition of the Mg2Si-based TE materials, the processing techniques, the micro- and nanostructures formed, and the final material properties of relevance to TE applications.

On the material processing side, a number of methods were used to achieve 3D bulk nanocrystalline Mg2Si and more than 700 samples were manufactured, analysed and tested. Notwithstanding the difficulties encountered due to the significantly different mechanical behaviour of Mg and Si, Mg2Si-based bulk materials could be successfully manufactured using silicon nanopowder. However, the desired improvement of TE properties with nanostructuring resulted being constrained by the limitations on the percentage of nanoparticles achievable in Mg2Si-based materials to be sufficiently dense for TE applications, and the final nanostructured products showed an overall degradation of the thermoelectric properties. These experimental results were also predicted by theoretical studies carried out in ThermoMag to study the influence of nanostructuring and grain boundary scattering on the TE properties of Mg2Si-based materials. Furthermore, experimental results seem also to confirm that upon aging the nano/micro structure of the Mg2Si-based samples tends to evolve into larger crystallites, which is an additional detrimental factor because, to be effective, nanostructured features should be maintained when actually integrated into a thermoelectric generator (TEG) for real-life applications. Notwithstanding these limitations, the ThermoMag project achieved outstanding results synthesizing n-type products which revealed a world record performance for n-type Mg2Si-based materials. Unfortunately, compared to n-type results, only moderate ZT could be achieved experimentally with p-type Mg2Si materials, which also found an explanation based on theoretical studies performed within the project. Consequently, MnSi1.75-based silicides (HMS) were selected as alternative p-type material for manufacturing ThermoMag TE modules because structurally, thermally and chemically stable at elevated temperatures and because of their potential for further improvement of thermoelectric performance.

Different types of demonstrator TE modules were designed with the help of computer modelling, confirming promising perspectives for real life applications of the materials developed. Nearly 1,000 n-type Mg2Si- and p-type MnSi1.75-based TE elements were manufactured and integrated into the demonstrator modules based on ThermoMag design for subsequent testing. Test results confirm expectations, clearly showing their high potential for further improvements. However, more work is needed for investigating further the silicide materials developed and to consolidate and improve the achieved results. Furthermore, more work on metallization and manufacturing processes is needed to improve the quality of the TE modules.

In addition to the scientific and technological activities, the ThermoMag project pursued also “Environmental” activities pursued via Health, Safety and Environmental assessment, Life-Cycle Analysis and Cost-Benefit Analysis. Clear evidence of ThermoMag overall impact in the field is provided by its 40+ publications. Project results are encouraging and show clearly that more emphasis is needed from the industry perspective in order to enable the successful commercialization of thermoelectric technology.

Project Context and Objectives:
As the global demand for energy increases, so does the demand for fossil fuels that are currently used in transportation, power-generating industries, factories and homes. The inevitable depletion of these resources lies at the root of a dramatic escalation of socio-economic and political unrest since decades already. On top of these growing concerns, there is also the environmental impact of climate change due to the combustion of fossil fuels which is becoming increasingly alarming.
These concerns on energy and environment matters have reached the highest profile at a global scale, and are well reflected into the efforts of the European Commission to promote and launch the Europe 2020 flagship initiative for a resource-efficient Europe, putting forward a series of long-term policy plans in areas such as transport, energy and climate change [EC,COM(2011)21]. The main target is therefore to develop and implement major energy-saving and CO2-reducing solutions.

In both the transportation and the manufacturing sector well over half of the energy consumed is discharged as thermal losses to the atmosphere or to cooling systems. These losses of energy are largely in the form of low pressure-low temperature waste heat which is not practical or economic to recover with turbine generators. Even if cannot be fully captured, these losses represent a major opportunity to help reducing our energy footprint through the scavenging of waste heat with thermoelectric generators. These are compact solid-state devices with no moving parts which convert temperature gradients directly into electrical energy (Seebeck effect). They are silent, reliable and scalable, making them ideal for distributed power generation and in applications where low maintenance need is crucial and/or no cooling fluids are allowed close to hot surfaces (e.g. in vicinity to liquid metals).

For real industrial applications of this technology, it is critical to focus on thermoelectric materials that are relatively cheap, widely available, non-toxic, lightweight, practicable and scalable for industry. In order to meet these very tight requirements, the core concept of the ThermoMag project revolves around developing and delivering new energy harvesting thermoelectric TE materials and modules, based on nanostructured bulk Mg2Si solid solutions. In fact, magnesium and silicon are both extremely abundant elements and hence very cheap compared to other conventional but rarer TE elements (Bi, Te, Ge, Sb etc.). Moreover, the ≈500°C peak temperature of Mg2Si-based materials is highly suitable for many areas of industrial waste heat recovery.

The main scientific objective of ThermoMag is to reveal a profound and quantified understanding of the links between the starting composition of the Mg2Si-based TE materials, the processing techniques, the micro- and nanostructures formed, and the final material properties of relevance to TE applications. In order to prove that the concept works, demonstrator modules are assembled integrating the energy-harvesting materials developed within the project. Furthermore, the real-world scalability of the thermoelectric materials investigated is assessed for automotive, aircraft and space applications. In addition to these scientific and technological objectives, the project includes also “environmental” objectives pursued via Health, Safety and Environmental assessment, Life-Cycle Analysis and Cost-Benefit Analysis.

Scientific Strategy for Improving ZT
The performance of thermoelectric materials depends on the so called “figure of merit” ZT=TS2σ/κ, where T, S, σ, and κ are the absolute temperature, Seebeck coefficient (or thermopower), electrical conductivity and total thermal conductivity, respectively. ZT can be increased by increasing S or σ, or by decreasing κ, which are all intrinsic properties of the specific material used. Unfortunately in conventional materials, it is rather difficult to improve ZT. Firstly, a simple increase of S will generally lead to a simultaneous decrease in σ. Moreover, an increase in σ leads to a comparable increase in the electronic contribution to κ. An alternative way to increase ZT is to reduce the thermal conductivity κ without affecting the electronic properties. Also, deliberately creating impurity-induced band-structure distortions could be a viable route to enhance S and ZT in all TE materials (doping). In this vein, a number of scientific strategies are investigated in this project:
• Produce, sinter and retain nano-sized grains in order to increase boundary scattering of phonons and hence reduce lattice thermal conductivity;
• Produce nanocomposites with the addition of inert nano-sized semiconductor inclusions/particles, again to reduce thermal conductivity;
• Doping and co-doping with small quantities of additional elements to alter band structures;
• Exploiting the unique number of natural isotopes in the Mg2Si system (i.e. 24Mg, 25Mg, 26Mg, and 28Si, 29Si, 30Si) and hence preferentially doping with higher mass isotopes.

With respect to measuring thermoelectric transport properties (S, σ, κ, as a function of T), standardised round-robin testing between labs are carried out to confirm the reproducibility of test results. In addition to the usual transport properties, other material features are also studied including mechanical and corrosion properties, annealing behaviour, thermal stability/cycling, ageing and re-crystallisation of nanostructures using for example HR-TEM.

Proof-of-Concept Testing
In order to prove the concept works, demonstrator TE modules are assembled that integrate the new energy-harvesting materials developed. The module design takes into account the aspect ratio of thermoelements (legs), the choice of ceramic supports, electrical/thermal connections and their optimisation, diffusion barriers, and possible segmentation/functional-gradation if required. Interface resistance, especially electrically, is an issue when dealing with nano-TE material and this is investigated using metallised coating of particles and atmosphere control to alter barrier height. Optimisation of the heat exchange is performed with the help of thermal modelling. All the above methods lead to the lab-scale production of (nanostructured) n- and p-type legs for material testing and for manufacturing TE modules. The proof-of-concept modules produced are then tested for investigating their overall performance.

Assessment of Industrial Applications
The strategy to focus on low-cost, high-ZT thermoelectric materials, made from easily available elements like Mg and Si, permits very widespread application of thermoelectric generators and coolers. In this project, the end-users identify a number of products where this technology could play an important role, demanding also additional functional requirements that need to be taken on board when developing a new class of TE materials and modules.

Health, Safety & Environment
While it is well known that bulk magnesium, silicon and Mg2Si are all non-toxic to humans and animals, the principle of precaution is adopted when handling nano-Mg2Si. Therefore, ThermoMag work includes also an “health, safety and environmental (HSE) assessment” of the possible risks and hazards associated with nanostructured materials based on Mg2Si, taking into account Material Safety Data Sheets (MSDSs) for the starting materials and dopants, workforce protective measures, the safety and confinement of nanoprocessing techniques, transport packaging for nanomaterials, training courses for those handling nanomaterials, and the assessment of REACH and RoHS obligations for nanomaterials.

Project Results:
Early project activities:
At project start, the relevant requirements related to the use of nanostructured TE materials and modules for automotive, aeronautic and space applications were captured by the project end-users who agreed on a common set of specifications for TE modules that would satisfy the majority of their needs and technical requirements.
Another matter that needed to be addressed at a very early stage of the project is related to the accuracy of measurements, which is an important and rather complex matter in thermoelectricity. For this reason, the ThermoMag partners dedicated special attention to standardisation issues and carried out two extensive inter laboratory (round-robin) test campaigns which showed that all the different transport properties measurements performed within the project fall within the uncertainty of the experiments (Fig.1).
Finally, before starting the research activities, Health & Safety (H&S) issues related to the working with and handling of nanostructured materials were addressed too. This resulted in a report collecting H&S assessment, relevant literature and a “Laboratory Safety Guide”. The project partners attended also a training course on H&S specifically focused on aspects of nanomaterial processing and handling. Therefore, all project work in ThermoMag has been fully compliant with local and EU regulations related to safe nano-production, safe handling and worker protection.

Materials processing, nanostructuring and characterization:
On the material processing side of the project, many ThermoMag partners produced Mg2Si and its derivatives by a number of methods, including different routes to prepare the doped Mg2Si starting material, namely low-cost combustion synthesis, mechanical alloying and high-temperature solid-state synthesis in inert crucibles. Various ball milling (BM) approaches were used to produce doped Mg2Si nanoparticle constituents. Systematic studies were performed to assess the effect of ball milling parameters on the mean particle size and particle size distribution of Mg2Si, demonstrating that adequately selected milling conditions allow producing powders with mean particle size of less than 2 μm and crystallite size going down to about 10 nm. Investigations provided also experimental evidence that a critical issue to secure completion of the synthesis of Mg2Si is to optimise the particle size of Si powder. Alternative methods were investigated too, including the feasibility of powder faceting by sonication in liquid media The thermoelectric constituents in powder form can be compressed into bulk via rapid spark plasma sintering (SPS) or hot pressing (HP) in vacuum. The temperature/pressure/time conditions were adjusted to the chemical composition of the powders in each technique. Other techniques, such as hot isostatic pressing (HIP) or explosive compaction for densification of undoped Mg2Si were tested too. In some procedures long-duration homogenising heat treatment was used at the intermediate stages or at the final stage of the manufacturing process.

The material characterization procedures at consecutive manufacturing stages comprised phase composition, chemical composition, microstructure and density (XRD, SEM/EDS, TEM/EDS). Reaction progress was studied by DTA/TG and DSC. Unique (nano)characterization was performed running X-ray synchrotron experiments at the European Synchrotron Radiation Facility (ESRF) and neutron diffraction experiments at the Institut Laue-Langevin (ILL) to determine 3D nanostructure, phase content, crystallography and thermal responses. Analyses revealed that in the manufacturing of ternary and quaternary materials, the products were often multiphase, consisting of solid solutions with different stoichiometry and intermetallic phases formed by dopants and components of the matrix.

Nanostructures were identified at different manufacturing stages: in ball-milled powders of reactants, in the agglomerated particles of reaction products, and in the sintered dense materials. However, the manufacturing of Mg2Si from silicon nanopowder in ThermoMag resulted very challenging at all stages from homogenizing and compaction of elemental powders to the sintering of the products. Due to the significantly different mechanical behaviour of Mg and Si, milling of the hard Si and the soft Mg particles results in a mixture composed of Si particles embedded in a Mg soft matrix. It is also observed that MgO impurities are always present in all ball-milled samples as well as in the commercial Mg2Si powder. With these studies, the ThermoMag partners demonstrated experimentally that the sintering of samples composed of Mg2Si nanoparticles is actually very problematic as the small size of the grains favours the oxidation of magnesium at grain boundaries and prevents high green densities which lead to non-cohesive pellets when sintered. Even when experiments were carried out under protected atmosphere during sintering, it was impossible to manufacture dense and cohesive pellets. It was therefore proposed to manufacture dense materials by mixing powders with different grain sizes. This approach has been successful, but the consequence is that the desired improvement of TE properties with nanostructuring is constrained by the limitation on the percentage of nanoparticles achievable in Mg2Si-based materials to be sufficiently dense for TE applications. Unfortunately, notwithstanding all these efforts, the final nanostructured products presented only slightly lower thermal conductivity and even increased electrical resistivity, showing an overall degradation of the thermoelectric properties.

Doping using various elements were predicted using electron band structure and ab-initio density-functional theory (KKR-CPA-type) modelling. The modelling results on more than 50 different dopants in Mg2Si obtained using electron band structure and ab-initio density-functional theory (KKR-CPA-type) predicted the most promising candidates for both n- and p-type doping for further experimental investigations. After a first round of experimental investigations, the n-type dopants chosen for use in the syntheses were Bi, Sb, and Co, while the p-type dopants were Ga, Ag, Li, Mo and B. Additional experimental investigations reduced further the number of potential dopants.
Among the investigated n-type dopants, bismuth and antimony were finally selected. First key results were achieved at an early stage already with the synthesis of good quality n-type Bi- and Sb-doped polycrystalline Mg2Si-based materials. Known good thermoelectric materials based on Mg2Si1-xSnx solid solutions at 0.4
Theoretical studies were carried out in ThermoMag to predict the influence of nanostructuring and grain boundary scattering on the TE properties of Mg2Si-based materials. Calculations carried out on phonon scattering at grain boundaries suggest an optimum grain size of about 50nm. However, the difficulties described earlier in achieving experimentally visible improvements of the TE properties via nanostructuring were confirmed by theoretical predictions carried out on nanostructured binary and ternary Mg2(Si,Sn) materials. These studies investigated the possibility of improving the thermoelectric properties of both Mg2Si and Mg2Si0.8Sn0.2 solid solutions via nanostructuring and grain boundary scattering, concluding that the solid solution Mg2Si0.8Sn0.2 is more favorable for the increase of the figure of merit due to nanostructuring than Mg2Si [D.A. Pshenai-Severin et al, J. Electron. Mat.,42,7(2013)]. However, assuming a hypothetical 100% nanostructured material, the achievable increase of the figure of merit due to grain boundary scattering with respect to the same microstructured material is rather modest and it is even less effective in the higher temperature range of interest for industrial applications (10% and 15% respectively for Mg2Si and Mg2Si0.8Sn0.2 at 850K). Moreover, these theoretical results were obtained assuming 100% nanostructured material but, due to the inevitable presence of regions inside the sample were the material is not nanostructured, the overall result will finally depend on the geometry and volume fraction of all phases. Considering that the ZT in a polycrystalline phase is lower than in the nanostructured phase, it can be concluded that, compared to the theoretical results for 100% nanostructured Mg2Si-based compounds, the overall effective ZT for the more realistic case of materials which include also phases with larger crystal sizes will be even smaller depending on its volume fraction. These conclusions seem to be confirmed also by work performed independently outside the ThermoMag consortium [N. Satyala and D. Vashaee, J. Appl. Phys. 112,093716(2012)], where the authors obtained similar results concluding that the enhancement of ZT due to nanostructuring in Mg2Si is problematic because the mean free paths of phonons and electrons are of the same order of magnitude.
In addition to the above, experiments were carried out to understand the long term effect of temperature on nanostructured Mg2Si0.98Bi0.02 bulk samples subjected to ageing by keeping its edges at elevated temperature (600K) for a prolonged time while the middle was kept at moderate temperature (310K). These experiments showed that nano-crystals were better retained in the centre than in the aged edges of the sample (Fig.3) suggesting that upon aging the nano/micro structure of the Mg2Si-based samples tends to evolve into larger crystallites, which is an additional detrimental factor because, to be effective, nanostructured features should be maintained when actually integrated into a TEG for real-life applications.

The project results summarized above clearly indicate that there are serious limitations to the possibility of improving the TE properties of Mg2Si-based materials with nanostructuring. However, notwithstanding these limitations, the ThermoMag partners managed to achieve outstanding results synthesizing n-type materials with composition Mg2(Si0.55Sn0.4Ge0.05):Bi by solid state reaction followed by hot-pressing, which revealed a very high figure of merit of ZT≈1.4 at around 800K (Fig.4) [A.U. Khan et al, Scripta Materialia,69,8(2013)]. These outstanding results were reproduced and also confirmed by measurements taken at different labs. Moreover, long term thermal stability tests carried out during more than 300 hours under temperature gradient from 320K (cold side) to 800K (hot side) in pure argon atmosphere evidenced no significant degradation of their good thermoelectric performance under the measurement conditions (Fig.5) therefore confirming the excellent properties of this n-type alloy [A.T. Burkov et al, Materials Today, in press]. Further investigations are needed to confirm these encouraging results and work is presently still on-going to further optimize phase concentrations and synthesis/manufacturing procedures.

For what concerns p-type, only Ga and Ag were experimentally confirmed among the theoretically predicted dopants for Mg2Si-based materials, although the ZT values achieved were less than half compared to n-type (max ZT<0.4 see Fig.6). Moreover, silver doping showed also rather ambiguous behaviour upon thermal cycling, seemingly due to the Ag displacements as a result of varying solubility of silver in Mg2Si and/or MgAg, as well as thermodynamic stability of phases in the Mg-Si-Ag system. One of the possible means to prevent these instability problems due to the formation of unwanted intermetallics is seen in ex situ doping, i.e. doping at the densification stage, and first experiments confirmed the feasibility of this approach for Ag, Bi and Sb. It was also demonstrated that the extension of sintering time up to 60 min in SPS has only minor influence on the grain size of the final product, which is of interest for further developments of this ex situ doping approach.
The ThermoMag partners worked throughout the whole course of the project on testing alternative dopants, synthesis, sintering and characterization of ternary and quaternary materials produced by different routes. More than 700 samples were manufactured, analysed and tested within the project. As discussed above, the best results were achieved with Mg2(Si,Sn,Ge)-based n-type materials doped with Bi or Sb which were selected for the construction and testing of thermoelectric TE modules, while research confirmed the difficulties in achieving comparable good results with Mg2Si-based p-type materials. These experimental findings showing high ZT for n-type and modest ZT for p-type Mg2Si-based materials seem to find an explanation in the results of theoretical studies performed within ThermoMag to study the influence of the relativistic effects on the electronic band structure and the thermopower of Mg2X semiconductors (X = Si, Ge, Sn) which show detrimental consequences for the thermoelectric performance of p-doped Mg2X, while similar calculations show negligible effect for n-doped Mg2X [K. Kutorasinski et al, Physical Review B,89,115205(2014)]. Similarly, ab-initio studies carried out elsewhere independently from ThermoMag on the defect and phase stability of Mg2X solid solutions (with X= Si, Ge or Sn) concluded that, due to compensation effects decreasing their thermoelectric performance when doped, p-type Mg2Si materials would actually be particularly difficult to obtain [R. Viennois et al., Intermetallics, 31(2012) and R. Viennois et al, AIP Conf. Proc. 1449(2012)].

Therefore, while efforts to investigate Mg2Si-based p-type materials continued to elucidate the behaviour of Ag (p-type) dopant, higher manganese silicides (HMS) were selected as alternative p-type materials for manufacturing the TE modules because structurally, thermally and chemically stable at elevated temperatures. Moreover, even though also the HMS used for manufacturing ThermoMag TE modules exhibit moderate thermoelectric figure of merit just above 0.4 at 800K (Fig.7) this p-type material offers better prospects when compared to p-type Mg2Si-based materials. In fact, in addition to its better stability, it is recognized that there is still room for improving HMS, especially by decreasing the thermal conductivity with different methods. Therefore, based on electronic structure calculations, a number of dopants were proposed in ThermoMag to enhance HMS performance experimentally, and nanostructuring techniques were investigated as well. Work to improve HMS performance is still ongoing and preliminary results confirm ZT improvements up to 0.55 at 800K for undoped HMS (Fig.8).

For what concerns the investigation of the effect of isotope doping on the thermoelectric properties of Mg2Si, lattice conductivity calculations performed in ThermoMag predicted that Mg2Si with isotopic Mg and Si could decrease the thermal conductivity up to 6% (Fig.9). Both undoped and Bi-doped isotopic samples were manufactured in order to investigate this effect experimentally on the following selected compositions: 24Mg26MgSi, Mg228Si30Si and 24Mg26Mg28Si30Si. For comparison, corresponding non isotopic reference samples were manufactured too and, being the expected effect rather low, it was first decided to study the effect of the purity of Si on the thermoelectric properties of these reference samples (Fig.10). As expected, the carrier concentrations of the Bi-doped samples resulted being higher than the undoped samples. Moreover, the carrier concentrations of the samples synthesized with Si 99,99% resulted being slightly higher than those of the samples synthesized with Si 99,9999%. Furthermore, since the uncertainty on the measurement of the specific heat capacity is large, results were compared making use of the measured thermal diffusivity instead of the thermal conductivity. These measurements showed that the thermal diffusivity of the sample synthesized with Si 99,99% is lower than the one of the samples synthesized with Si 99,9999% for the undoped samples whereas it is the opposite for the Bi-doped samples. A similar behaviour is observed also by plotting the thermal diffusivity divided by sample density (Fig.11) and, since the measurement of the specific heat capacity is not accurate enough to observe small variations in the thermal conductivity, further measurements will need to be carried out making use of a microcalorimeter.
With regard to the isotopic samples manufactured, initial results (Fig.12) show that the measured thermal diffusivity of the isotopic samples is higher compared to the reference samples, which is the opposite of what was expected. Further investigations are being carried out to complete the study of the effect of the impurities on the thermoelectric properties and to measure the other isotopic samples.
Results on isotopes achieved until now don’t seem to be particularly promising. However, investigations are still ongoing, and conclusive results shall be presented via future journal publications.

Proof-of-concept TE module manufacturing and testing:
The thermoelectric materials selected for integration into ThermoMag proof-of-concept TE modules consist of Mg2Si- and MnSi1.75-based (HMS) materials respectively for n- and p-type TE elements. However, the active materials represent only a small -even though essential- part of a thermoelectric device. Therefore, not only the thermoelectric properties but also the thermal stability, corrosion resistance and thermo mechanical behaviour must be investigated in order to validate the use of any material in a working device. Research on the n-type Mg2Si-based materials developed in ThermoMag indicate already that the thermoelectric performance of the optimized materials could be reproduced several times, validating the compositions and the manufacturing processes. These materials showed also rather stable thermoelectric properties upon thermal cycling, while corrosion tests clearly demonstrated the need to protect them from oxidation, especially at temperature above 500°C. As already discussed, these issues are less of a concern for HMS. Furthermore, TE materials and electrodes must be connected through the stacking of both a diffusion barrier, which must be sufficiently thick in order to avoid inter diffusion and electro-migration phenomena, and a metallization layer to insure good electric connection. It must be remarked also that a possible drawback of the chosen combination of n- and p-type materials is the large difference in thermal expansion coefficients (CTE) of Mg2Si- and MnSi1.75-based materials (Fig.13) which, due to the temperature gradients and variations involved, could induce the formation of cracks or the complete fracture of the TE module assembly. For this reason, solutions were developed to ensure the lifetime of the thermo-elements

Based on the known properties of the selected materials, computer simulations performed by various end-users showed that, even though at higher working temperatures, the expected power of the modules is in the same range as presently available with other high performance materials. These modelling results seem promising for real life applications, with the potential for achieving even higher power with future improvements of the silicide materials investigated.
Different types of TE modules were designed and nearly 1,000 TE elements (both n- and p-type) were manufactured by the ThermoMag consortium for their integration into TE modules and for subsequent testing (Fig.14). Work performed included the preparation of the materials, the coating of both TE elements and electrical contacts and their soldering.
Investigations on interfacial materials to be used in all-silicide TE modules were carried out, enriching the overall knowledge of this complex and crucial stage of module manufacturing. The n- and p-type legs were sputtered with different metallization layers (Fig.15) to find suitable material stacking for both hot and cold side. Both Cu and Mo electrodes were used for the hot side, combined with Cu electrodes on the cold side. The Cu/Mo electrodes were also sputtered with materials like nickel, lead and chromium. Layers were approximately 4µm for both leg types and showed good adhesion to the TE legs (Fig.16). However, it should be remarked that the use of Pb, which is prohibited due to its toxicity, and the use of Cu electrodes, which finally demonstrated to be problematic for high temperature applications mostly due to the fast diffusion of Cu, should not be retained for the future.
Further investigations indicated Ni as potentially suitable interfacial material at the hot side, particularly for n-type Mg2Si legs. Nickel foils or powders, also combined with a buffer layer of intermediate mixtures of magnesium silicide and nickel powders to offset CTE mismatches, were added below and on top of the thermoelectric material prior to the densification with HP or SPS. Using this approach, TE elements with Ni electrodes could be obtained directly by dicing these pellets with in situ manufactured nickel contacts (Fig.17). This seems unfortunately to be less promising for the metallization of the p-type HMS legs, as the CTE mismatch between Ni and HMS generates mechanical stresses that impair their robustness. However, for HMS metallization, first studies indicated TiSi2 as a promising material, even though further characterization and reproducibility tests are needed to confirm these findings.

Prior to the assembly of the TE modules, a risk assessment was performed to address risks that could have raised and hampered the timely and satisfactory achievement of the project objectives. Conclusions highlighted contacting and coating related risks as well as mechanical stability risks due to the different CTEs of the n- and p-type materials used to for TE module manufacturing.

Several test-modules were produced using various combinations of ThermoMag Mg2Si-based n-type and MnSi1.75-based p-type TE elements (Fig.18). Used for technology testing, the design of these modules adopts a system with pressure contacts on the hot side which reduces the occurrence of mechanical failures induced by the different thermal expansion of the materials used. This design allowed also easy reconfiguration for testing different materials and metallization solutions.
These TE modules performed well during testing (Fig.19) starting with low inner resistance. Test results show how the open circuit voltage reached 0.65V at 405ºC hot-side temperature and 25ºC cold-side temperature, the actual temperature drop over the module being somewhat smaller around 350ºC (Fig.20). At this temperature gradient a maximum power of 1.04W was measured, yielding 0.28W/cm2 per module area, or 0.98W/cm2 per area of the thermoelements only (~0.6W/g per weight). The inner resistance was measured at several temperature gradients and it was found to be very stable at around 0.1Ω. The performance was degrading only slowly at steady-state testing. However, the performance and the inner resistance degraded quickly during thermal cycling tests.

In addition to these test modules, cylindrical modules with similar pressure contact system as well as various fully-soldered flat modules without pressure contact system were manufactured too (Fig.21). Compared to the test modules adopting similar spring load system for pressure contacts, preliminary test results on the cylindrical module showed better stability and performance at similar temperature gradients (Fig.22). For what concerns fully-soldered flat modules, initial performance tests indicate performances in line with the designs with spring load system. However, due to the absence of the system with pressure contacts on the hot side, problems of cracks or even mechanical failure of the whole assemblies arose as a consequence of the different coefficient of thermal expansion of the materials used.

From the tests performed on the various ThermoMag TE modules series presented above it can be concluded that the results achieved with the Mg2Si-based n-type and the MnSi1.75-based p-type materials developed by the ThermoMag consortium are very promising, clearly showing their high potential for further improvements. However, more work is needed to investigate further these materials and to consolidate and improve the achieved results as the measured absolute performances are still rather low compared to TE modules based on other present state-of-the-art materials. Furthermore, the achievement of good thermal and electrical contacts, good reliability and the prevention of mechanical failures/cracks in the materials remain major challenges. Further work on metallization and manufacturing processes is therefore needed to improve the quality of the TE modules developed.

Industrial and environmental Analyses:
Industrial and environmental analyses were carried out within the project through a supply chain analysis (SCA) for the Mg2Si- and MnSi1.75-based TE materials developed in ThermoMag (dopants included), as well through life-cycle (LCA) and cost-benefit analyses (CBA) by taking in account the selected target applications for passengers cars, trucks and aircrafts. These analyses indicate that there should be concern over long term dependency on some of the materials used like antimony, bismuth and germanium, while there is less concern in the global supply of silicon, magnesium and manganese. The supply and market analyses of the materials enabled also to determine the cost of ThermoMag TE modules from a materials’ perspective. This showed, for example, the impact on cost of dopant materials like germanium, providing an incentive to find alternatives and cheaper substitutes. Another key aspect to be considered for future developments will be the increased reliance on recycling of materials from scrap sources and also from the recycling of thermoelectric modules. With this in mind, recyclability trials were carried out aiming at manufacturing Mg2Si using Si recovered from slurry waste produced during wire saw cutting of Si-PV cells. Disassembly and interconnect studies on TE modules manufactured within the project were performed too.
Analyses of the raw-materials and production processes for the thermoelectric elements developed in ThermoMag focussing on the scaling-up to industrial mass production quantities concluded that the production methods for the TE silicide materials developed by ThermoMag labs may all be scaled up to an industrial scale. Several technological developments are proposed too, including different approaches for production up-scaling to accelerate the synthesis/sintering of reaction products, to manufacture contact layers directly in-situ during densification as well as to manufacture pellets of larger diameter via SPS. Further technology developments on the preparation and deposition methods should aim at enhancing the manufacturability of devices, increasing their efficiency and lowering degradation speed.
An important conclusion of all these analyses performed in ThermoMag is that future efforts at European R&D level should focus more on TE modules’ development optimization and on their cost reduction. This is based on the observation that the cost share of TE modules manufacturing is by far the main driver of the overall cost of a TEG, not the cost of the raw thermoelectric materials used (Fig.23). Fundamental topics such as high temperature working conditions, stability in thermal cycling, elements’ diffusion, differential thermal expansion, elements sublimation etc. must be further investigated and optimised considering low cost processes as the primary driving factor. All these aspects are of fundamental importance to the path of thermoelectric device design, assembly, integration and testing that could ultimately lead to scale-up production and commercialization.

These results are summarized in the ThermoMag “European Research Roadmap” which provides an analysis of the state of the art of thermoelectric energy harvesting, recognizing present main achievements/weak points and main challenges and aiming at identifying the real and concrete future needs and the steps forward necessary to overcome present limitations. In particular, the ThermoMag “European Research Roadmap” stresses the importance of closely coordinating the research efforts aimed at developing sustainable TE materials with those aimed at developing TE modules and generators.

Potential Impact:
Drivers for the interest in energy harvesting via the direct recovery of waste heat and its conversion into useful electrical energy with thermoelectricity are based on achieving better efficiencies in thermal engines and thereby reducing CO2 emissions and the focus of ThermoMag has been mainly on recovering waste heat from transport vehicles and factories because of their major scaling-up potentials. For these applications, materials choice is not only driven by the intrinsic TE properties but it is also influenced by other important factors like non-toxicity, availability, durability and cost. The same applies for the TE modules, where not only the conversion efficiency is important but also other factors like simple/straightforward design and manufacturing are decisive for scalability. For space application the focus is clearly placed on mission-enabling energy technology, while for aircrafts it is mainly placed on energy harvesting to enable wireless sensor and actuator systems.

The European research on thermoelectricity has been particularly active in the last decade, especially on material research, and the ThermoMag project provided its valuable contribution with promising developments on sustainable silicide materials. However, it is also recognized that more emphasis is needed from the industry perspective in order to enable the successful commercialization of thermoelectric technology. In fact, the ultimate efficiency of TE devices is usually limited by technological issues, like heat transfer, mechanical properties, phase stability, and/or electrical and thermal contact resistance between materials and electrodes. Furthermore, it is also demonstrated that at present conditions the raw TE materials cost is negligible with respect to TE module cost, while the latter holds also the largest share of the total cost of a thermoelectric generator. Moreover, it is also recognized that the performances obtainable with some of the existing materials with realistic sustainable scale-up potential (i.e. ZT values sufficiently close to unity in the medium-high temperature range, as it is the case for ThermoMag) are already sufficient for thermoelectric devices to become products for large scale markets.

These prospects for industrial applications of thermoelectric technology are generating worldwide increased activity in this field, demanding materials with sustainable scale-up potential. Therefore, while continued research in advanced sustainable thermoelectric materials is necessary and recommended, the future research should focus more on low cost synthesis processes, on the development of efficient and mechanically robust TE materials and devices as well as on their scalability. More work is needed to research, develop, validate, and demonstrate advanced heat transfer materials, thermal interface materials, heat exchanger technologies and protective coatings to counteract long-term in-service degradation of TE materials and devices. All these aspects are of fundamental importance to the path of thermoelectric device design, assembly, integration and testing that could ultimately lead to scale-up production and commercialization.

To conclude, the development of efficient, thermoelectric devices using only low-cost, readily available, renewable and sustainable materials would certainly contribute to address a range of major challenges impacting on energy efficiency, climate change and resources depletion. In order to achieve its ultimate potential in industrial waste energy recovery it is therefore imperative to closely coordinate the research efforts aimed at developing sustainable TE materials with those aimed at developing TE modules and generators.

The significant results achieved by ThermoMag are largely demonstrated by the number of publications which so far include 31 peer reviewed publications in high-impact international journals (additional 5 publications submitted recently), 7 publications in international conference proceedings, scientific monograph or edited books/book series, as well as extended abstracts and more than 20 posters presented at various conferences worldwide.
In a nutshell, ThermoMag has resulted in:
(i) ZT up to 1.4 for n-type doped Mg2Si-based material achieved (ZT≈0.4 for p-type);
(ii) manufacturing of μ- and nanostructured Mg2Si-based material with different techniques;
(iii) good reproducibility of TE measurements confirmed with two round-robin campaigns;
(iv) modelling on nanoscale mechanisms of more than 50 dopants for material design;
(v) advanced material characterization;
(vi) health & safety assessment and training;
(vii) a common set of specifications for TE modules based on end-users’ needs;
(viii) design, manufacturing and testing of a variety of TE modules using project’s own materials;
(ix) industrial, environmental and scalability analyses developed by industrial partners;
(x) 40+ high quality publications and other dissemination activities;
(xi) IPR training, patent search and patent applications;
(xii) a “European Research Roadmap” outlining future R&D directions in thermoelectrics.

The list of all dissemination activities throughout the whole course of the project is impressive, so far consisting of more than 150 entries. These include, among others, two video lectures on thermoelectricity which are now available online via the ThermoMag website, YouTube and ESA web streaming as well as an exhibition on thermoelectricity at the Leonardo da Vinci Science Museum in Milano (Italy). Furthermore, two very successful European Conferences on Thermoelectrics were organized and chaired by ThermoMag partners during the course of the project (Fig.24). Other dissemination activities include workshops, lectures, flyers, presentations, trade fair appearances, interviews, TV clips, etc.

The overall volume and quality of the dissemination efforts from all ThermoMag partners clearly demonstrates the impact of the project in disseminating non-confidential information to the scientific, industrial and educational communities and the general public. Moreover, both the number of publications as well of all other dissemination activities is expected to increase even further in the coming months.

List of Websites: