Final Report Summary - MAG-DRIVE (New permanent magnets for electric-vehicle drive applications)
Executive Summary:
Electro-motors may replace internal-combustion engines in the future vehicles. It is therefore of a vital importance to invest in development of electric drive. The working principle of any electric motor is based on a magnetic-torque-induced rotation, hence it is indispensable to provide a magnetic field, which can be produced either by coils or by permanent magnets. The formers make us possible to continuously control the field up to extremely high values by varying the applied electric current, whereas the magnets operate without any external power and can be shaped almost arbitrarily to fit into any motor design. Modern high-performance permanent magnets are made of alloys, which contain rare-earth elements. Due to the economical and political situation in the world the availability of these materials is limited. Particularly severe is the situation in the case of the so-called heavy-rare-earth elements, like dysprosium and terbium, which are added to the basic materials to enhance the performance of the magnets up to the level defined by the specifications of the electric motor. In order to reduce the manufacturing costs and even to prevent unexpected delays in production, it is necessary to invest in development of corresponding materials with a reduced amount of heavy-rare-earth elements, which is the main objective of the MAG-DRIVE project.
Further requirements for the material are related to the specific working conditions and environment: the temperatures of about 100 0C and corrosive atmosphere, in which the electric motors installed in vehicles operate. The next objective is therefore to make a magnet not just without using heavy-rare-earth elements and with the specified performance, but also with an ability to perform at such conditions. And the final goal is to integrate the magnets into an electric motor and/or an alternator with an improved and specially developed thermal management and power electronics, suitable as a power aggregate for electric vehicles.
The performance of a magnet is characterized by its hysteresis loop, which represents the response on an external magnetic field. The area surrounded by the loop is related to the energy stored in an operating magnet, hence it should be as large as possible. Consequently the value of the external field, required to demagnetize the magnet, called coercivity, and the output of the magnet in the absence of the external field, called the remanence have to be high enough. Additionally, the squareness of the hysteresis loop impacts the surrounded area too. Although these parameters depend on the intrinsic properties of the starting material, they can be substantially influenced by the processing of the powders and the way how the powders are consolidated to form solid magnets. Whereas the remanence can be controlled by the density and alignment of the compacts, the standard way to provide a high coercivity used to be based on adding heavy rare-earth elements to the starting powder materials. An alternative is to keep the size of the grains within the particles as small as possible. There are several ways to achieve this goal. The most obvious seems to provide the powders with sufficiently small particles. In this manner we approached along two different routes, either by means of the so-called wire-explosion procedure, where the fine powder is formed upon applying a high-electric-current pulse, or using various sophisticated techniques. Since the chemical composition of the powder obtained from the wire explosion does not match the on the hard-magnetic phase it was necessary to compactify the powders in terms of the Kochanek process, which made us possible to modify the composition by keeping the particle size low. The powders resulting from milling were sintered in a standard way after carefully tuning the respective temperatures, heating rates and durations. Another option to deal with the small grains is in the frame of the so-called HDDR process, which stands for the hydrogen decrepitation, deabsorption, recombination. Basically, it is a method where hydrogen gas induces a decay of the material, which then reforms with an improved morphology. The so-obtained powders were finally exposed to the spark-plasma sintering (SPS) to make the solid magnets.
Within the MagDrive project we investigated the three processing routes, which at the end all led to to the demonstrator magnets with desired properties. However, it turned out that the magnets obtained from the HDDR+SPS route had the best characteristics, therefore the demonstrator electric-motor was built with them. The motor passed all required tests and could be in principle applicable in real vehicles.
The results in the investigations within the MagDrive projects were presented in several peer-reviewed publications and on some major scientific conferences. We applied for one patent related to the details of processing. In order to demonstrate that the developed procedure is not important just from the scientific point of view, we proved its economic importance by preparing a financial plan for the exploitation of the HDDR+SPS procedure for a massive production of magnets. It is expected that the break-even point, at which he initial one-off investment costs equal the profit, would occur at the end of the third quarter of the third year since the start of the production.
Project Context and Objectives:
The MAG-DRIVE consortium consisted of research partners: Jozef Stefan Institute (Slovenia), University of Birmingham (United Kingdom), Institute for Chemistry, Technology and Metallurgy (Serbia), Queen Mary University of London (United Kingdom), of a research and development company (SME) Kochanek Entwicklungsgesellschaft (Germany), of a magnet-producing company Magneti Ljubljana d.d. (Slovenia), and of a producer of components for automotive industry Valeo (France). Jozef Stefan Institute (JSI) is the leading Slovenian research institute with a lot of experiences in production and characterization of magnetic materials by means of magnetic measurements and electron microscopy. JSI was the coordinator of the project, contributed to all activities with an emphasize on the magnetic measurements and electron microscopy. The University of Birmingham is ranked in the top 100 universities worldwide. The Magnetic Materials Group in the School of Metallurgy and Materials has a reputation as one of the leading groups in the field, focused on development of novel processing techniques, which was also their main role in the project, combined with magnetic measurements and material characterization. The University of Belgrade's Institute of Chemistry, Technology and Metallurgy (ICTM) is leading institution in the field of micro-system and nano-system technologies in the Western Balkan. They mainly performed mechanical milling and characterization of powders. Queen Mary, University of London is one of the UK's leading research focused higher education institutions. The Functional Nanomaterials Group has particular expertise in processing including the spark-plasma sintering, which was their contribution to the project. Kochanek Entwicklungsgeselschaft is a small privately held SME, specialized in radical new concepts for manufacturing nano-crystalline metal parts, some of them were applied in the project. Magneti Ljubljana d.d. is privately owned SME for making permanent magnets. They provided conventional sintering and some magnetic measurements. VALEO is an independent industrial group fully focused on the design, production and sale of components, integrated systems and modules for the automotive industry. They prepared and performed all sorts of tests of the demonstrator magnets, and built the electro-motor with the required power electronics.
The description of work was divided into nine work-packages (WPs).
The WP1: Project Management lasted through the whole course of the project, and it was about the project management, and it was completely covered by the Jozef Stefan Institute. The main goals were to:
Ensure efficient project coordination, progress monitoring, budget and financial control
Ensure the quality of all activities and the timely delivery of reports and deliverables
Ensure effective and on-time communication between the project consortium and the European Commission through the REA
Prepare the rolling 6-monthly Action Plan
The activities were divided into the following tasks:
Task 1.1: Project management (M1 – M36)
Task 1.2: Preparation of 6-monthly Action Plan (M1 – M30)
Task 1.3: Creation and maintenance of project website (M1 – M36)
The results, entirely contributed by the Jozef Stefan Institute, were presented in 3 deliverables.
The WP2: Technical requirement lasted from the month 1 to the month 8, and it was led by Valeo. It was devoted to a detailed survey of industrial specifications and technical requirements for the demonstrator and the new electric motor, as described in the following tasks:
Task 2.1: Benchmarking electric motors (M1 – M3)
Task 2.2: Design of thermal management for power electronics (M3 – M8)
Task 2.3: Assessment of currently available permanent-magnet (M1 - M8)
The results, contributed by all partners, were presented in 4 deliverables.
The development of material was covered in the work-packages WP3, WP4 and WP5.
The WP3: Route I-The oxide route was led by Kochanek Entwicklungsgesellschaft, and it lasted from the month 1 to the month 30. The primary objective was to produce nano-crystalline hard-magnetic material by applying the wire-explosion experiment and the Kochanek process. The investigations comprised the following tasks:
Task 3.1: Preparation of test cores from rare-earth oxides (M1 – M12)
Task 3.2: Preparation of hard magnetic materials (M9 – M18)
Task 3.3: Preparation of complex 3D-shaped permanent magnets for industrial use in Evs (M15 – M30)
Task 3.4: Development of theory for hard magnets based on the Kochanek process (M18 - M30)
The results, contributed by Kochanek Entwicklungsgesellschaft, Jozef Stefan Institute and University of Birmingham, were presented in 6 deliverables.
The WP4: Route II-The nanoparticle route was led by the Institute for Chemistry, Technology and Metallurgy, and it lasted from the month 1 to the month 30. The main goal was to obtain magnetic powder with the particle size of the order of magnitude of 1μm by means of innovative milling techniques through the following tasks:
Task 4.1: Analysis of the nano-milling technologies (M1 - M6)
Task 4.2: “Light-gas” milling and comparison studies (M6 – M20)
Task 4.3: Separation of magnetic particles (M12 – M24)
Task 4.4: Production of Dy-diffused Nd–Fe–B magnets (M20 – M30)
The results, contributed by the Institute for Chemistry, Technology and Metallurgy, Jozef Stefan Institute, Valeo and University of Birmingham, were presented in 5 deliverables.
The WP5: Route III – The HDDR+SPS route lasted from the month 1 to the month 30, and it was led by the University of Birmingham. The goal was to increase the grain size by applying the hydrogen-decrepitation-deabsorpition-recombination (HDDR) process in order to achieve high coercivity, and to spark-plasma sintering (SPS) to make the magnets as described by the following tasks:
Task 5.1: HDDR processing conditions (M3 – M15)
Task 5.2: Alignment and pressing of green compacts (M9 – M18)
Task 5.3: Investigation of densification kinetics during SPS (M12 – M24)
Task 5.4: Spark-plasma sintering of aligned green compacts (M18 – M30)
The results, contributed by University of Birmingham, Jozef Stefan Institute, the Institute for Chemistry, Technology and Metallurgy and the Queen Mary, University of London, were presented in 6 deliverables.
The WP6: Microstructural and magnetic characterisation lasted from the month 3 to the month 35, and led by the Jozef Stefan Institute, provided the service for WP3, WP4 and WP5 with constantly ongoing activities by means of analysing the structural and magnetic properties of the processed materials. The tasks were:
Task 6.1: Scanning electron microscopy (M3 – M35)
Task 6.2: Advanced electron microscopy (M3 – M35)
Task 6.3: Magnetic measurement (M3 – M35)
Task 6.4: Particle size and surface analyses (M6 – M30)
Task 6.5: Corrosion tests (M12 – M35)
The results, contributed by Jozef Stefan Institute, Valeo, Kochanek Entwicklungsgesellschaf, University of Birmingham, the Institute for Chemistry, Technology and Metallurgy and the Queen Mary, University of London, were presented in 18 deliverables.
The WP7- Production of permanent magnets and Thermal Management Components was led by Magneti and Valeo, and it lasted from the month 12 to the month 30. The aim was to make the magnets from the powders prepared in the processing WP's, and to construct the power electronics for controlling the electric motor. The activities were described in the following tasks:
Task 7.1: Preparation of magnet-processing (M12 – M20)
Task 7.2: Pressing, sintering and magnetization of samples (M14 – M24)
Task 7.3: Assessment of permanent magnets (M20 – M26)
Task 7.4: Thermal management of power electronics (M18 – M28)
Task 7.5: Optimisation of EM power density using new permanent magnets (M24 – M30)
The results, contributed by all partners, were presented in 5 deliverables.
The WP8: Testing of the Magnets and the Electric Motors in the Real-World Conditions, led by Valeo, and lasted from the month 24 to the month 35, represented the final research work package, since the output was a working electric motor, based on the MagDrive magnets. The work was divided into the tasks:
Task 8.1: Preparation for the tests (M24 – M26)
Task 8.2: Physical and magnetic properties of the permanent magnets (M26 – M28)
Task 8.3: Corrosion properties of the permanent magnets (M26 – M28)
Task 8.4: Tests on the new electric motor (M28 – M34)
The results, contributed by Valeo, Jozef Stefan Institute and Magneti, were presented in 5 deliverables.
The WP9: Dissemination and exploitation, was led by the Jozef Stefan Institute and lasted through the entire course of the project. It was devoted to the dissemination and exploitation of the MagDrive results in various forms according to the proposed tasks:
Task 9.1: Dissemination (M1 – M36)
Task 9.2: Exploitation (M1 – M36)
The results, contributed by all partners, were presented in 3 deliverables.
The objectives of the project were valid during the entire duration of the project with differing intensity.
A: The focus of the overall objectives was targeted towards:
developing high-coercivity hard-magnetic materials without heavy-rare-earth elements like dysprosium or terbium by dramatically reducing the grain size to below 1 μm using three novel processing routes
using the developed materials to produce demonstrator magnets applicable for electric-vehicle motors
optimisation and the thermal management of the power electronics and rotors in electric motors
efficient project coordination, quality of all activities, timely delivery of reports and deliverables, effective and on-time communication between the project consortium and European commission, preparation of rolling 6-monthly Action plan
B: Objectives related to short-term activities and progress:
to provide industrial specifications and the technical requirements for the magnet demonstrator and the new electric motor, which was completely achieved in the WP2
to build up nanocrystalline hard-magnetic materials in principles of the wire-explosion experiment, which was completely achieved in the WP3.
to provide magnetic powder of particle size in the order of magnitude of 1 μm by means of “light-gas” milling procedures, which was completely achieved in the WP4.
to produce HDDR-processed Nd-Fe-B anisotropic magnetic powders with the grain sizes in the sub-micron region, which was completely achieved in the WP5.
to constantly provide a characterisation of structural and magnetic properties of the materials, produced in WP's 3,4 and 5 by means of electron microscopy and state-of-the-art magnetic measurements, which was completely achieved in the WP6.
to prepare the magnet-processing equipment, and to perform the pressing, sintering and magnetization of samples, which was completely achieved in the WP7.
to define and to report on the testing procedures, and to report on the structural, magnetic and corrosion-resistance tests, which was completely achieved in the WP 8.
to maintain external project communication, and to guarantee the impact of the project through the creation of a comprehensive exploitation plan, which was completely achieved in the WP9.
The flow-chart of the project included four different milestones:
MS1: Technical Specification for New Magnets defined within the WP2, led by Valeo was expected to be achieved by the end of the month 8
MS2: Midterm Review defined within the WP1, led by the JSI was expected to be achieved by the end of the month 18
MS3: First Demonstrator Assembled defined within the WP7 led by Magneti was expected to be achieved by the end of the month 23
MS4: Project Finished defined within all WP's led by the JSI was expected to be achieved by the end of the month 36
Project Results:
The S&T results were obtained in the work-packages WP2, WP3, WP4, WP5, WP6, WP7 and WP8. Within the WP2 we set the technical requirements for the demonstrators. The WP3, WP4 and WP5 represented three different routes towards the same goal to produce magnetic powders with the specified properties. The powders were consolidated in the form of solid magnets within the WP7, which was also devoted to the construction of the power electronics, required to control the electric motor, build with the magnets within the WP9. The WP6 was horizontal, and it covered all magnetic and structural characterization, required in other research-and-development-oriented WP's.
WP 2: Technical Requirements (M1 - M8)
WP Leader: Jean-Marc Dubus (VALEO)
Summary of Results
All activities within the WP 2 were finished in the first eight months. According to the plan we performed a detailed survey of industrial specifications and technical requirements for the demonstrator and the new electric motor. It was found that there were two competitive technologies in terms of performance and price, both with some advantages and disadvantages. Available products from the leading manufacturers were analysed in detail. The problem of cooling was addressed not just by designing but already by providing some prototypes for the required electronic circuits. Finally, we obtained the information about the currently available permanent magnets to be build in the electric motors.
There were no deviations from the objectives and tasks given in Annex I.
Details for each Task
Task 2.1: Benchmarking electric motors (M1 – M3)
Electrical vehicles can be configured in a number of different ways. Based on various considerations, for example, required output power and the available space. A choice was made between integrated starter generators (ISGs) and belt-drive starter generators (BSGs):
The main advantage of the ISG machine is that it can be used for ZEV driving and to maximize the regenerative braking power, as the thermal engine could be disconnect through a clutch mechanism.The main drawback of the ISG machine is the mechanical integration in between the combustion engine and the gearbox, which increases the operating costs.The comparative analysis of the competitor machines focused on the following: power density, efficiency, high torque at low speed in E-drive, cooling system, use of raw materials (in particular the use of RE), machine technology
We tested machines from the following companies:
Honda
Porsche
Mercedes
Cadillac
Lexus
Toyota
Kia
These motors where purchased on the open market from the OEMs. We tested some electrical motors that fall into the following categories:
Mild hybrid electrical vehicle (HEV)
Mild & Full HEV
Full Hybrid with Plug-in
The motors were tested as received and then stripped and cut away for imaging and other tests on the materials in the motors. Detailed tests were carried out on motors from:
Porsche (manufactured by Bosch)
Toyota
Renault (manufactured by Continental)
VW (manufactured by ZF Sachs)
The motors ranged in terms of the power output from 20 kW to 45 kW. Sizes were in the range 200 to 300 mm. Special attention was given to the position and orientation of the magnets. The Toyota motors all used embedded magnets with a “V” configuration. The German motors favoured a surface positioning of the magnets. It is important to note that the Continental motor, produced for the Renault Kangoo contains no magnets. Of course this motor has a large weight penalty as a result. The best power density was achieved by the Toyota Yaris electrical machine. This is mainly due to the use of high-grade Nd-Fe-B magnets and a high stator-filling ratio of 2,5kW/kg of raw material. The lowest power density was measured for the Renault Kangoo motor due to the wound rotor: 1kW/kg.
The main advantages of belt-driven machines are that they fit in the place of standard alternator, they are cheap and could achieved 85% of the CO2 reduction at about half of the price of an ISG system. However, the main drawback is that they are not as efficient as the ISG architecture in regenerative-braking mode. In the comparative analysis of our competitors we focused on the following: power density, efficiency, high torque at low speed in E-drive, cooling system, use of raw materials (in particular rare earths), machine technology.
In the MAG-DRIVE BSG investigation we looked four belt-driven devices. The electrical machines were produced by Hitachi, Bosch and Mobis. These motors where purchased on the open market from the OEMs. The rated power ranged from a low 5 kW to 82 kW, 32 kg Bosch SMG 180. These machines were all about 20cm long.
Task 2.2: Design of thermal management for power electronics (M3 – M8)
Under the consideration were two types of the water-cooled inverters which can operate at the temperatures up to 80 0C: the 24 V B2-9 related to the crankshaft motor generator and the 270 V charger-inverter type CMG B2-6. Both Devices were developed by VALEO.
Crankshaft Motor Generator B2-9:
Voltage range 36/55 Vdc
Typical voltage 48Vdc
Maximum phase AC current 350Arms
Temperature range -40°C / 80°C Water cooled
270V Charger inverter specification CMG B2-6
Typical voltage 270Vdc
Voltage range 220/300 V dc
Maximum phase AC current 260Arms
Temperature range -40°C / 80°C Water cooled
Task 2.3: Assessment of currently available permanent-magnet (M1 - M8)
RE-Fe-B-based magnets were taken from some of the motors and sent for a full chemical analysis. We noticed a remarkably high Dy content in excess of 10 wt.%. This is approximately twice as high as we were expecting. The quantities of the other elements were in line with expectations.
WP 3: Route I – The Oxide Route (M1 - M30)
WP Leader: Wolfgang Kochanek (KE)
Summary of Results
The primary objective of the WP 3 was to produce nanocrystalline hard-magnetic material by applying the Kochanek process which is based on the reduction of oxides. The results of extensive experimental investigations revealed that samarium oxide could not be reduced in any form at any conditions, hence the conclusion was drawn that the samarium-cobalt magnets could not been made by means of the oxygen-reduction process. It was also demonstrated that the reduction of neodymium oxide was relatively low, therefore an additional infiltration of the neodymium nanoparticles is required in order to get the optimum results. It was decided to achieve this goal by performing the so-called wire-explosion experiments, where the required nanoparticles are formed upon the applying an extremely high electrical voltage between the two ends of a ribbon made of respective material. The resulting pulse current yields a decay of the ribbon into nanoparticles. The required ribbons were produced by spark-plasma sintering. Finally, we applied the Kochanek process to prepare the demonstrator magnets. The analysis, based on the scanning-electron microscopy and magnetic measurements, proved the route, developed in the frame of the WP3, promising for a massive production of permanent magnets, although further improvements in a reduction of the oxygen content, enhancement of the density, and of the amount of the hard NdFeB phase in the magnets are still required.
There were no deviations from the objectives and task given in Annex I.
Details for each Task
Task 3.1: Preparation of test cores from rare-earth oxides (M1 - M12)
The Kochanek process is based on the reduction of metal oxides. Therefore the feed stock for the Sm2Co17 consists of Sm2O3 and Co3O4 with organic compounds and of the Nd2O3, Fe3O4 and B with organic compounds in the case of the Nd2Fe14B. The main challenge consists in the reduction of the rare-earth oxides which have a high thermodynamic stability. The Gibbs free energy are 1085,3 kJ/mol and 1009,6 kJ/mol at room temperature for samarium and neodymium, respectively, indicating that the reduction under this conditions is thermodynamically not favourable With rising temperature the Gibbs free energies drop only a little. At 1800 K they still reach 844,5 kJ/mol for samarium and 858,6 kJ/mol for neodymium which does not significantly improve the thermodynamically conditions for reduction and suggests the equilibrium to be far on the left side of the reaction.
In the case of samarium a series of experiments was conducted which produced a total of over 40 samarium oxide based samples. The researched parameters in these reductions include temperature in a range from 650°C up to 1360°C, reduction duration from 3 h up to 6 h, precursor density from 3.29 g/cm3 up to 4,19 g/cm3 and different compositions of additives. Regardless of all the variations of parameters the weight loss measurements clearly proved that no samarium oxide reduction could be achieved as demonstated by means of the XRD spectroscopy. Hence, we conclude that no samarium oxide can be reduced in order to apply the Kochanek process for making the Sm2Co17 magnets.
In the case of neodymium a series of experiments was conducted which produced a total of over 250 neodymium oxide based samples. The researched parameters in these reductions include temperature in a range from 650°C up to 1360°C, reduction duration from 3 h up to 9 h, precursor density from 1,47 g/cm3 up to 3,49 g/cm3, precursor particle size from roughly around 100 nm up to 5 μm, homogeneity of precursor material varied through the preparation process and different compositions of additives. Depending on these parameters a neodymium oxide turnover rate from 0% up to 20% could be achieved. The turnover rate was determined through the weight loss of the samples during reduction process. Simple by hand prepared samples achieved if at all only a low turnover rate. The highest neodymium oxide reduction rate could be reached through application of the more complex Kochanek process which yielded smaller particles and a higher homogeneity. At high temperature and long duration these samples reached a turnover rate close to 20%. The sample consisted of Nd2O3, Fe3O4, B and organic additives and was prepared according to the Kochanek process. The starting density was 3.11g/cm3 with a height of 6.06 mm and a diameter of 13.00 mm. After reductive treatment at 800 °C for 4 h a complete iron oxide turnover was achieved but none of the neodymium oxide. The density was lowered to 2.45 g/cm3with a height of 6.17 mm an a diameter of 12.63 mm. After pressing with 4.5 t/cm2 the sample showed a density of 4.47 g/cm3 with a height of 3.15mm and a diameter of 13.00 mm. A secondary reduction at 1360°C for 5 h yielded a neodymium oxide turnover rate of 8.6% determined through weight loss and a final density of 6.21 g/cm3 with a height of 2.86 mm and a diameter of 11.5 mm. Compared to the theoretical density of 7.65 g/cm3 the material reached 81.2%. The XRD analysis of the produced samples reflect the results obtained through weight loss measurements. Even though XRD does not allow to determine an exact neodymium oxide turnover rate the found phases depend on the composition and thus on the turnover rate. Samples treated with lower temperature whose weight loss measurements show no or little neodymium oxide turnover rate, show only iron and neodymium oxide phases in their corresponding XRDs. XRD results for materials prepared along the Kochanek process with high reduction temperature and a neodymium oxide turnover rate close to 20% showed beside a main neodymium oxide and iron phase a small pure Nd phase. The EDS analysis showed a significant higher neodymium oxide turnover rate for oxides close to the surface. It was found that iron rich areas as well as iron lean areas achieved a similar neodymium vs. oxide ratio of about 7 : 3. This corresponds to a neodymium oxide reduction rate of 71% for oxides close to surface. The magnetic analysis of the samples show according to the previous results only a soft magnetic response due to the high iron content. The samples showed no coercivity and therefore no hard magnetic phases have formed.
We conclude that the reduction of neodymium oxide in the presents of Fe/Fe3O4 could be achieved up to 20% indicating that the thermodynamic stability is different in the presence of iron compared to pure neodymium oxide. To achieve a competitive hard magnetic NdFeB-material a neodymium oxide turnover rate close to 100% is needed since small impurities in the material have a negative influence on the magnetic properties. Trying to push the equilibrium of the reduction by high temperature is limited as we found out that the matrix starts to melt beyond1400°C. Which is supported by the iron neodymium phase diagram. Therefore we proposed an additional infiltration of the neodymium nanoparticles by performing the so-called wire-explosion experiments. The nano-sized powder was formed upon applying an extremely high electrical voltage between the two ends of a ribbon made of respective material. The resulting pulse current yielded a decay of the ribbon into nanoparticles. The required ribbons were produced by melt spinning.
Task 3.2: Preparation of hard magnetic materials (M9 – M18)
We used super-structured co-precipitated Nd2O3 /Fe3O4 material and conducted a series of experiments which produced a total of over 60 super-structured co-precipitated samples. The experiments where conducted in a temperature range from 700°C up to 1300°C and a duration of reduction of 4h. Samples chemical reduced with a temperature lower than 900°C re-oxidized on air at room temperature. The lower the reduction temperature the more rigorous the samples re-oxidized. This well known effect is based on the lesser sintering at lower temperatures resulting in a more reactive part. It is noticeable for the co-precipitated materials that the re-oxidation occurs at reduction temperatures up to 900°C. Comparable samples based on mere oxide mixtures did not re-oxidise at the researched temperatures with the lowest temperature being 600°C. This shows a more fine-grained particle size distribution for the co-precipitated materials. A SEM-imaging comparison between a neodymium oxide powder and the co-precipitated material illustrates the difference in particle size and therefore the reactivity due to a higher surface area. While the neodymium oxide particles size was above one micron the particle size of the co-precipitated material was within the sub-micron range. Regardless of all the temperature variations and the higher reactivity based on smaller particles the weight loss measurement results of all super-structured co-precipitated precursors showed a significant lower turnover rate instead of an expected increase in turnover. These losses in turnover rate can be explained through the XRD results. Instead of forming a Nd2Fe14B phase the reduction led to a Fe main phase, a unexpected NdFeO3 phase, a Nd2O2S phase based on some sulphate impurities and a small pure Nd phase. Due to the formation of a NdFeO3 phase only an incomplete iron oxide reduction was achieved. This explains the significant lower turnover rate compared to samples based on oxide mixtures which in general achieve a close to 100% iron oxide reduction.
Task 3.3: Preparation of complex 3D-shaped permanent magnets for industrial use in Evs (M15 – M30)
During a wire explosion the “wire” reaches a temperature of more than 104 K with a heating rate of around 107 K/s. In general the explosion is achieved by the discharge of a capacitor through the conductive wire. Their capacity ranges from a fraction of a μF up to several hundred μF with storages voltages varying from a few kV to several hundred kV. Current peaks up to 50kA are common. We monitored the voltage and current time dependence for a wire electrical explosion of a steel wire done with the experimental setup built for the MAG-DRIVE project. The capacitor was charged with 6kV and while discharging the current reached a maximum peak of almost 8kA. The whole process took less than 300μs.The explosions took place in a special reactor, build for the purposes of the MAG-DRIVE project. The explosion chamber was affixed inside a steel frame. Screwed on the whole outside of the steel frame were 3 mm thick steel plates to shield against electro magnetic waves. The explosion chamber was rendered inert with an constant Argon stream. Also hooked to the chamber was a pump that circulates the contained gas. In between the circulating gas stream was a washing flask to collect the product. The size of the resulting particles was determined by means of the scanning-electron microscopy yielding the assessment that the main fraction had a diameter between 20 and 150 nm with an occasional deviations up to 2 μm. The low density iron part for the infiltration with Nd based nanoparticles was produced according to the Kochanek process. The starting density of the disc shaped iron Kochanek parts was in the range of 2.72 to 2.98 g/cm3, which equals a free pore volume of 61.4% to 58.8%. For infiltration a well stirred suspension of stabilized neodymium based nanoparticles was added drop-wise onto the Kochanek parts. To minimize the exposure to air the procedure was done in an argon ventilated glove box. The dropping speed of the suspension was adjusted to be as fast as the infiltration into the Kochanek sample itself. Dropping was carried out for as long as the suspension was sucked in.
The concentration of nanoparticles in the suspension needs to be 27,87% (v/v) for a free volume of 61.4% and 31.06% (v/v) for a free volume of 58.8% to be infiltrated in one session. Infiltration tests showed that suspension with such high concentrations of particles have a poor infiltration behavior. The parts will not get infiltrated thoroughly. Because of that a nanoparticle concentration of 15% (v/v) was used, which showed a good infiltration behavior up to the centre of the part in the preliminary tests.
The use of low concentration slurries made it necessary to infiltrate repeatedly. In between infiltration sessions the parts were dried until the cyclohexane evaporated. The procedure was repeated until the calculated amount of suspension was infiltrated into the Kochanek part.
After infiltration the samples were heat treated to take the organic components out of the part on the one hand and on the other hand to form the NdFeB hard magnetic phase. To evaporate the amines and to decompose the remaining organic components the samples were heat treated between 400 °C and 500 °C for several hours under hydrogen. Afterwards they were heated up to 850 °C for a duration of 10 hours under nitrogen to form the NdFeB hard magnetic phase. The parts reached a density from 3.85 g/cm3 up to 4.26 g/cm3.
Task 3.4: Development of theory for hard magnets based on the Kochanek process (M18 - M30)
During the course of the project it was established that the reduction of oxygen might not be the most efficient procedure for the production of magnetic powders containing rare-earth and transition-metal elements. Therefore we successfully applied an equally innovative alternative by means of the wire-explosion experiment. However, we were still able to explain the reasons for a failure of the initially-proposed process in terms of a plausible theory.
The idea was to reduce neodymim oxide by exposing the samples to the elevated temperatures in the range from 650°C up to 1400°C in the H2, N2 and/or Ar atmosphere. At temperatures above 1000°C the reduction of neodymium oxide starts to begin. Unexpectedly at reduction temperatures of 1200°C and higher the turnover rates caves in not even reaching a complete iron oxide reduction anymore. This decrease in turnover rate was accompanied with a melting of the sample surface. The molten surface could also be the reason for the low turnover rate, since it could be responsible for inhibiting the reduction by blocking the gas transfer between the centre of the sample and the surrounding atmosphere. An EDS analysis of the sample surface visualized the molten surface as dark grey areas that mainly consisted of iron. Even though the XRD and EDS results verified that a neodymium oxide reduction could be achieved to some degree there was no hint of a Nd2Fe14B hard magnetic phase in these results. Magnetic measurements of the samples confirmed these previous results as only a soft magnetic response, due to the high iron content, could be measured.
WP 4: Route II: The Nanoparticle route (M1 - M30)
WP Leader: Dana Vasiljević-Radović (ICMT)
Summary of Progress towards Objectives
The main objective of the WP4 was to obtain magnetic powder with the particle size of the order of magnitude of 1μm by means of innovative milling techniques. Towards this aim we analysed the commercially available equipment and decided for the most suitable technology which would serve as the starting point for further improvements. The application of this technology gave the satisfactory results in terms of the particle size close to the micrometer region. We quantitative analysed the size distribution and separated the powder.
There were no deviations from the objectives and task given in Annex I.
Details for each Task
Task 4.1: Analysis of the nano-milling technologies (M1 - M6)
The state of the art of different milling techniques that can be used to obtain sub-micrometer-sized nano-crystalline rare earth-transition metals magnetic materials was surveyed. Nanocrystalline NdFeB magnetic materials were considered since they possess the highest energy product of all the permanent magnets available today. We reviewed properties of different mechanical milling methods, including impact milling, ball milling, jet milling, roller milling, etc. We put the largest stress to the ball media milling and jet milling. The consideration includes the basic phenomena occurring in different milling procedures, the available equipment, including pros and cons for each type, different accompanying effects that may influence the quality of milling and the state of the art of experimental works in the field. Basic properties of NdFeB sub-micrometer and nano-powders were reviewed from the point of view of their influence on the production of permanent magnets. We performed also numerical simulation of different grinding and milling procedures by applying the the discrete element method. Based on the survey including the properties of both the powder material to be processed (brittleness, flammability, toxicity, etc) and the available techniques, we decided for the equipment most suitable for the use in our own experiments with NdFeB. It is a FRITSCH Planetary Micro Mill PULVERISETTE 7 fabricated by Fritch GmbH from Idar-Oberstein, Germany. The reasons for the decision are the following:
The grinding of materials down to the nano-meter range requires very high application of energy and therefore significantly higher rotational speeds than those allowed by typical planetary mills. Conventional planetary ball mills are characterized by grinding bowls that are clamped to the sun disk. This limits the maximum possible rotation speed because of a specific speed limit, the centrifugal forces acting on the bowls will be so great that the clamping of the bowls can no longer hold. Sinking of the grinding bowls into the sun disk of the mill solves these problems. In this construction, the centre of gravity of the bowls lies in the plane of the sun disk. The centrifugal force arising generate significantly lower overturning moment, which in turn allows the mill to run at significantly higher speed. As a result, this mill attains a speed of up to 1100 rpm, reaches centrifugal accelerations of up to 95 times the force of gravity, making the energy application approximately 150 % greater than conventional planetary mills. This significantly reduces the grinding time to reach the nano-meter range. For certain materials, as is NdFeB, brittle and hard, we think that only this level of energy application even allows the milling to NPs range. Grinding bowl size is 80 ml, ant it is made from tungsten carbide (WC). Grinding balls are from the same material. There are 2 bowls, max. sample quantity is 5ml and min. sample quantity is 0.5 ml. In bowls dry and wet grinding can be performed. We bought special emptying device which enables a quick and easy separation of the grinding balls and suspension. Grinding bowl is tightly sealed so that even grinding in suspension without any additional sealing is possible. Lead is equipped with pressure and temperature sensors and has a bleeder valve so that any overpressure in the bowl can be equalized in a controlled fashion. These features allow grinding in an inert atmosphere.
Task 4.2: “Light-gas” milling and comparison studies (M6 – M20)
The milling experiments were performed by the use of a JET-mill. The milling was accomplished by the use of light-gas milling of HD Nd-Fe-B material using a production-scale JET-mill. The comparative experiments were performed in the company Netzsch from Hanau in Germany, where they applied their ConJET dry grinding machine.
The starting material for the JET milling experiment was chosen to be a Magneti-standard NdFeB material with a low Dy content (1 wt. %). This material was chosen due to low intrinsic coercivity (HcI)- owing to the low Dy content. 50 kg of NdFeB flakes was HD processed for the experiment,
The smallest ConJet produced by Netzsch, ConJet 10, was used for the experiments. In contrast to a fluidized-bed Jet mill, ConJet does not have a fluidized milling chamber with a fluidisation nozzle on the bottom and a classifier on the top. The milling chamber is located around the classifier, with an increased number of nozzles. The milled powder particles path is therefore set longitudinal to the classifier, by which a sharper cut point of larger particles can be achieved compared to a fluidized-bed JET mill. In order to avoid oxidation of the incoming HD flakes and the collected (milled) powder, the whole mill was enclosed in a flexi glass bell, filled with inert atmosphere. The milling parameters were as follow:
• Milling gas N2, 99.95% purity (open circuit).
• Nozzle diameter: 1.2 mm.
• Nr. of nozzles: 5.
• Nozzle pressure 7 / 8 bar
• Classifier speed: 16,000 RPM/ 12,000 RPM / 3,000 RPM / 6,000 RPM / 9,000 RPM.
• Milling chamber pressure: 0.2–0.29 bar
The following two experiments were declared as a successful run:
• Ex1: Classifier: 16,000 RPM / p(nozzle): 8 bar / productivity: 0.2 kg/ h / D(observed) < 3 μm
• Ex2: Classifier: 6000 RPM / p(nozzle): 8 bar / productivity: 2 kg/ h / D(observed) < 8 μm
Task 4.3: Separation of magnetic particles (M12 – M24)
The first step in the magnetic-particles separation is the analysis of the powder size. We performed the on-site experiments with the SEI LOT Phenom PRO. The Ex1 (6,000 RPM / p(nozzle): 8 bar / productivity: 0.2 kg/ h) and the Ex2 ( 6000 RPM / p(nozzle): 8 bar / productivity: 2 kg/ h) materials were analysed. Although the Ex2 material obviously consisted of small particles, the productivity was too low, hence the average measured particle size was between 3 and 5 μm, which is very close to the submicron region, estimated as the optimum to achieve the desired magnetic properties.
Task 4.4: Production of Dy-diffused Nd–Fe–B magnets (M20 – M30)
The essential step was to separate the particles of the desired size from larger suspension volumes using a process combining sedimentation and membrane filtration. NdFeB particle suspension after surfactant-assisted wet milling is allowed to sediment for time t, then the upper portion of the suspension corresponding to height h is decanted and filtered in protective argon atmosphere under pressure ΔP through a membrane with a specific pore size. The process is repeated n times with the settled slurry and filtrate returning to another milling cycle. The filtration cake is composed of particles having a size distribution determined by sedimentation and filtration parameters. After filtration, polycarbonate membrane with the filtration cake consisting of NdFeB particles was immersed in 25 mL of chloroform to dissolve the membrane and disperse the particles. The resulting suspension was sonicated for 5 min and investigated with Zeta Sizer Nano ZS90 using a 1:10 dilution ratio. The resulting particles were characterized by the average diameter of 315 nm with most particles falling in the range from 200 to 500 nm. Such magnetic powder exhibits the desired properties, the only obstacle for an eventual massive production are relatively small quantities of the produced material, which calls for further investigations towards scaling up the involved procedures.
WP 5: Route III: The HDDR+SPS route (M1 - M30)
WP Leader: Allan Walton (UOB)
Summary of Results
Hydrogenation Disproportionation Desorption and Recombination (HDDR) is a high temperature hydrogen treatment that utilizes the ability of materials such as Nd-Fe-B to readily absorb and desorb hydrogen at elevated temperatures which results in a fine grained material with very interesting magnetic properties.
It was found that the optimal processing conditions for the material with a reasonably low Dy content were the temperature of about 880 °C, with a disproportionation pressure of 1500 mbar, recombination under vacuum and quick cooling upon completion. However, by increasing the Dy as well as the the Co content the required applied pressure should be increased in order to complete disproportionation.
In order to fully exploit the potential of the material, it is highly desirable to align the particles prior to the final compaction. The result of the conventional spark-plasma sintering is an isotropic green compact, therefore we developed a method to pre-align the starting HDDR powder in the presence of an external magnetic field as described in detail within the Task 5.2.
There were no deviations from the objectives and task given in Annex I.
Details for each Task
Task 5.1: HDDR processing conditions (M3 - M15)
In this task the processing conditions for each step of the HDDR process were optmized for a range of different Nd-Fe-B magnet compositions including primary cast materials and secondary scrap sintered materials. The cast material includes a range of ternary alloys that can be altered with alloying additions to optimize the HDDR reaction. A range of sintered magnets have been used with low Dy levels.
The aim of this task was to avoid over-processing of the Nd-Fe-B material during each step of the process without foregoing the anisotropy that can be developed in HDDR powders. The disproportionation step was monitored using mass-flow controllers and a thermocouple for each chosen composition. The impact of the processing temperature and hydrogen pressure has been assessed in terms of their ultimate value, ramp rate and hold time during the disproportionation and recombination reactions. The impact of each of these variables on the microstructure has been determined with the aim of minimizing the size of the Fe, Fe2B and the Nd hydride rods in particular, prior to desorption. In a similar fashion the pressure-reduction rate and quench rate for the desorption step was monitored in order to minimize grain growth on formation of the Nd2Fe14B phase during the recombination reaction. The magnetic properties of the HDDR powders were measured in WP6 to assess the level of anisotropy that was developed for each composition and set of processing parameters.
Three compositions of Nd-Fe-B magnets were chosen for the trials. Two sources from large sintered blocks of Composition A - Nd13.4Dy0.8Al0.7Nb0.3Fe78.5B6.3 (at %) and Composition B – Nd12.5Dy1.8Al0.9Nb0.6Co5.0Fe72.8B6.4 (at %). The first composition was very low in Dy as the target of this project is to lower the Dy content of fully dense sintered magnets. The second composition was chosen as the Dy content is slightly higher but still below the average Dy content measured across a range of mined sources (Wang 2013). Following on from this '′real′′ scrap magnets were processed based upon voice coil motor magnets from hard disk drives. These were chosen as hard disk drives represent the largest source of sintered Nd-Fe-B from consumer products, they are collected in large quantities for data removal and there is relatively little variation in composition from product to product.
The HDDR processing conditions have been optimized for the processing of 20g sintered Nd-Fe-B batches of Composition A (low Dy). This was achieved by changing the process temperature (830-930 °C), disproportionation pressure (1000-2000 mbar), recombination pressure (0-350 mbar) and recombination time (0-30 mins) in order to produce the greatest magnetic properties. The optimal process conditions for Composition A were found to be at 880 °C, with a disproportionation pressure of 1500 mbar, recombination under vacuum and quick cooling upon completion. The voice coil motor magnets with a similar composition also processed successfully within the optimized parameters. Composition B which contains higher Dy and Co content, however, required a higher hydrogen pressure in order to complete disproportionation. By increasing the disproportionation pressure to 2000 mbar as required by Composition B, samples of Composition A exhibited lower remanence and were less anisotropic due to over-processing.
Having tracked the disproportionation reaction of the various sintered, book mould and strip cast starting alloys, it could be observed that the reaction kinetics were strongly affected by composition, rare earth content and initial microstructure. It was found that for the cast starting materials, the microstructure had the largest effect on the disproportionation reaction when compared to rare earth content. Strip cast alloys initiate disproportionation at lower hydrogen pressures than conventional book mould cast alloys which could be attributed to the much smaller grains in the strip cast material allowing for fast propagation of the disproportionation reaction through the matrix grains. Increasing neodymium content in ternary alloys increases the rate of reaction for hydrogen decrepitation but delays the onset of disproportionation. This can be attributed to the excess of Nd-rich grain boundary phase which reacts readily during hydrogen decrepitation but acts as a transport medium for hydrogen before disproportionation initiates within the matrix grains. For sintered magnets the additions of Co and Dy also delay the onset of the disproportionation, requiring an increased hydrogen pressure to initiate the reaction.
Task 5.2: Alignment and pressing of green compacts (M9 - M18)
For the production of aligned green compacts suitable for spark-plasma sintering (SPS) compaction, anisotropic HDDR powder was produced. The HDDR powder was a mixture of coarse and fine particles when removed from the processing furnace, and so it required breaking down using a pestle and mortar to remove any agglomeration. After 5-10 minutes of light grinding, the powder was suitable for green-compact production. 15 g of the HDDR powder was then loaded into a neoprene isostatic press bag and filled to tap density. A rubber bung was then pushed into the top of the press bag to a tight fit and then sealed using electrical tape. The seal was to ensure that liquid could not be introduced to the sample during the pressing stage and also prevents oxidation when the sample is prepared under an inert atmosphere. The sealed press bag was then placed inside the coil set from a capacitor-discharge pulse magnetiser. This piece of equipment consists of a large bank of capacitors that when pulsed from a fully charged state produce a 9 T field through the copper coils in the direction from ceiling to floor. The sample was situated centrally within the coil and held in place using a wooden-support system to ensure that the press bag is kept completely vertical. If the press bag was to lean to the side or be allowed to move then the alignment of the powder within the press bag would not be perfect. The capacitors were charged and discharged three times to pulse the powder particles three times. This allowed for full alignment of the powder particles in the upwards direction. The press bag containing the aligned powder sample was then placed inside the water-filled chamber of the isostatic press. The top ram was then inserted into the chamber and the pressure increased to 10 tons and held for 1 minute. The pressure was then very slowly released to prevent cracking of the sample as the walls of the neoprene isostatic press bag released from the surface of the sample. Once removed from the isostatic press chamber, the green compacts were ejected from the press bag and inspected. The green compact does not have a parallel flat top and bottom faces. This is due to a combination of the particle size of the starting powder, the magnetic field exhibited by the compact and the lack of a binding agent as used in bonded magnets. The powder particles in these regions follow the magnetic field lines of the bulk of the magnet, this reduces the density of the upper and lower regions of the green compact. There are a number of ways in which this problem could be addressed. Mechanical removal of the ‘fluffy’ end parts using a blade is possible; however, the overall alignment of the sample could be reduced during the cutting process as the powder particles are disturbed. The large variation in particle size could be impacting on the stacking of the particles during compression. By reducing the particle size it could be possible to produce better compaction under isostatic pressing.
When reducing particle size it was important to consider an appropriate milling technique. Wet milling techniques such as low-energy roller ball milling requires cyclohexane carrier fluid and milling times from 1-20 hours to produce a fine particle size. The powder would then become very fine, leading to an increased potential for oxidation and yield loss during the drying process in a vacuum port. Jet milling would produce a very fine powder with particle sizes close to 1 μm and below. These particles would be very susceptible to oxidation and unfortunately due to the low sample size (20g) jet milling is out of the scope of this work.
The milling process chosen for this work was a laboratory sized hammer mill. This milling technique utilises three rotating blade arms that grind the powder against a ring of teeth that line the internal wall of the mill. A 0.2mm hole size mesh is placed in the bottom of the mill for the milled powder to pass through and into a collection tube. The built-in speed control unit allows for careful control of the milling process to avoid sparking of the Nd–Fe–B during the grinding of the particles between the rotating arms and the integrated grinding teeth.
The hammer mill was run for 10 minutes to allow all of the HDDR powder to pass through the 0.2 mm mesh. Due to the depth of the teeth in the hammer mill, the HDDR powder became caught during the milling process, reducing the yield collected in the tube; however, blowing compressed air through the loading hopper released the powder from the teeth to continue the milling process. The milled HDDR powder was then passed through the isostatic pressing route described above. The reduced particle size allowed for better stacking of the powder and hence a more uniform green compact was produced. There was a small amount of fine particles remaining in the bottom of the press bag after the green compact was removed, however they did not form along the magnetic field lines, as in the previous samples. However, due to the lack of a binding agent and the nature of the pressing route the powder around the edge walls of the surface became loose as the compact was handled; therefore, an alternative pressing technique was also considered.
Uniaxial pressing using a 20-mm die set allows for a higher pressure to be used, as well as controlled specimen dimensions due to the polished internal wall of the die set and the upper and lower punch pieces. As with the conventional SPS compaction route, this preparation method produces isotropic samples as there is no current alignment field. However, the pressing force used with this press and die set is 20 tons, rather than the 0.5 ton compaction used with the graphite die sets. It can be observed that the green compact is very uniform in size and shape; however, as previously stated the magnet is isotropic in nature.
A possible solution to this is that a copper wire could be wound around the die set and an electrical current be passed through the wire during the pressing action to create a magnetic alignment field within the die set. Unlike the isostatic alignment route, where the powder is pulsed aligned before pressing, this route would maintain a constant alignment field throughout the entire pressing process; however, the alignment field would be much weaker.
A potential improvement for the isostatic pressing route would be to powder blend the HDDR powder with milled neodymium hydride powder. The hydride powder would serve two roles in this process, firstly due to the powder being soft and friable, it would potentially acts as a binding agent to keep the green compact intact during and after removal from the isostatic pressing bag. Also it would aid with liquid-phase sintering during the SPS process, improving the densification behaviour of the compact.
Task 5.3: Investigation of densification kinetics during SPS (M12 – M24)
The initial experiments were performed on large batches (400 g) of the low-Dy material with the composition Nd13.4Dy0.8Al0.7Nb0.3Fe78.5B6.3 since the main target of the project is to reduce the rare-earth content in the fully-dense sintered magnets. The magnetic properties, including anisotropy were reduced compared to the 20 g batches due to the highly endothermic nature of the recombination reaction. The temperature of the sample can drop by up to 100 °C during recombination, hence taking the sample out of the optimal processing window, leading to a loss in anisotropy. The grain size was submicron however; hence it was a good material for initial SPS testing i.e. subsequent grain growth during SPS treatment could be checked.
The green compacts for the SPS treatment were formed by lightly grinding the HDDR-treated sample with a pestle and mortar. The powder was then loaded into a neoprene isostatic press bag and sealed with a rubber bung and electrical tape. The bag was then placed into a capacitor discharge pulse magnetiser and pulsed three times with a 9T alignment field. The sample was then pressed to 10 tons using an isostatic press. The green compact was then removed from the press bag and wrapped with a single layer of graphite paper and pressed into a graphite die. The die was then loaded into the SPS machine which was sealed and evacuated.
A range of SPS treatment conditions were trialled, and the best results were found after processing at 900 °C for 1 minute, the sample was cooled under vacuum and ejected from the graphite die. The density of this sample was calculated as 7.14 g cm-3, which was measured with a densitometer. The theoretical maximum density for Nd-Fe- B is ~7.5 g cm-3, so the sample was ~95% dense. The microstructure of the samples were investigated by means of the SEM technique. We observed a mixture of very fine grains along with some large grains throughout the material. The average grain size was measured using the Feret diameter, and was calculated to be 0.75 μm with a standard deviation of 0.41 μm.
Task 5.4: Spark-plasma sintering of aligned green compacts (M18 - M30)
Like the conventional sintering, the spark-plasma sintering (SPS) is used to bind particles of material together into strong useable objects. This occurs through mass transfer at the atomic scale, acting to build ‘necks’ between particles. In addition to heating, pressure and electric fields are present during SPS to achieve the goal. The relatively high temperatures and sintering durations required to fully consolidate powders using conventional sintering techniques render the process unsuitable for the sintering of certain materials. For some materials, the sintering temperatures necessary to achieve full density are too high to sinter using conventional methods. In others, unwanted chemical reactions may occur during heating for long periods of time; and, where it is desirable for the properties or certain characteristics of the starting powder to be transferred to the final sintered compact, quicker processing times and/or lower processing temperatures are necessary to avoid any change in microstructure. For this reason, much of modern advanced materials sintering involves the use of Field Asisted Sintering Techniques (FAST). In contrast to the slow external heating methods employed in conventional sintering, FAST techniques employ the use of an applied electric current to produce high heating rates of up to 2000 K/min. Such heating rates are achieved through the application of pulsed current (from a few micro seconds to a few milli seconds) of extremely high current density - up to several kA. This is possible due to the high electrical conductivity of the tooling involved and is therefore achieved without the need for high voltages (typically below 10 V). Powders and green compacts are usually sintered within a conductive die. In the case of an electrically conductive green body, energy is dissipated directly within the sample and the conductive die; whilst for electrically insulating green parts, the heat is generated through the Joule heating of the die and is transmitted by conduction to the powder. The process yields a densification (shrinkage) of the exposed material. We applied the heating rate of 100 °C min-1 up to 900 °C and pressure of 50 MPa for four different powders.The fall of the rate of densification to 0 beyond 730 °C suggests that densification has completed. After sintering, the sample density was measured using Archimedes principle and confirmed to be near theoretical, hence applicable for the production of magnets.
WP 6: Microstructural and magnetic characterisation (M3 - M35)
WP Leader: Matej Komelj (JSI)
Summary of Results
This work package provided the service for WP3, WP4 and WP5 with constantly ongoing activities by means of analysing the structural and magnetic properties of the processed materials. In term of structural characterization the basic tool was the scanning electron microscope which makes us possible to qualitatively observe the microstructure, i.e. the phase composition, resulting from different processing conditions. By applying this technique we demonstrated the strength of the HDDR method in reducing the amount of the undesired soft-magnetic alpha-Fe phase. Advanced microscopy methods made us possible to quantitatively characterize the chemical compositions of the considered materials. Other techniques, like high-resolution transmission electron-microscopy (HRTEM) or laser confocal microscopy were used for imaging on atomistic level or of complex 3D shapes, which was important for the spark-plasma sintered compacts as presented in the description of the Task 6.2. Of equal importance as the microscopy were magnetic measurements. We relied on two different principles: the open-looped vibrating-sample magnetometer and the closed-loop permeameter. The former was more suitable for the samples in a powder form, whereas the latter was applicable for solid magnets. Both techniques were widely used throughout during the execution of the project. The particle size and surface morphology were of a great importance for the WP4. Different milling conditions lead to different particle sizes. It was very important to protect the magnets for electrical-vehicle applications against corrosion. An appropriate protection can drastically reduce a decay of a magnet due to environmental conditions as demonstrated in the Task 6.5.
There were no deviations from the objectives and task given in Annex I.
Details for each Task
Task 6.1: Scanning electron microscopy (M3 – M35)
Scanning electron microscopy is the single-most important analysing technique for materials science . The length scales involved at every stage of the procedure, from hundreds of microns to hundreds of nanometres, are well covered by this technique. The key information that was supplied here by the SEM technique relates to grain size and phase distribution. Of particular importance were the observations relating to book-mould-cast and strip-cast material. The SEM revealed the striking differences in microstructures that can be obtained with these techniques. Book-mould casting tended to produce larger grains, a coarser microstructure and more separated phases – and most importantly the presence of free iron, which is potentially at least very problematic for permanent-magnet production of any kind. The SEM studies revealed how strip casting, to a large extent, overcomes these problems.
By way of benchmarking the SEM technique was also applied to sintered magnets, where it was able to show that sintered Nd-Fe-B magnets also consist of two main distinguishable phases: Nd2Fe14B matrix grains surrounded by a network of Nd-rich triple junctions and grain boundaries. When SEM is used to look at HDDR material, which is material on a sub-micrometre scale it shows very nicely that the HDDR process is successful in ‘removing’ the α-Fe from the starting microstructure. When the HDDR process is applied to strip cast material the SEM is able to show that the long, columnar grains are also absent from the microstructure and that this HDDR processed strip cast Nd-Fe-B could lead to better compaction using spark plasma sintering (SPS) as the regions of cavitation are not as significant as in HDDR processed book mould cast Nd-Fe-B. The SEM was also used to look at SPS samples of HDDR powder. This starting powder was found to have an ideal submicron grain size distribution, typical of HDDR-processed material. However, there was also evidence of explosive grain growth in some of these samples, particularly those processed at high temperatures. Again, SEM is an ideal technique for assessing materials in this way and provides a clear explanation for the lower-than-expected coercivity values, but with the advantage of providing us with a clear path to overcome the problem. In the case of the best samples achieved so far for HDDR powder and SPS, based on grain size alone, the retention of a proportion of fine grains which are of single-domain size in the best samples suggests that the magnetic properties of these samples can be improved.
Task 6.2: Advanced electron microscopy (M3 - M35)
Advanced electron microscopy includes techniques like energy-dispersive X-ray spectroscopy (EDXS), electron-energy loss spectroscopy (EELS), high-resolution transmission electron microscopy (HRTEM), Lorenz microscopy, Laser assisted 3D atom probe (3DAP) microscopy, and Laser confocal microscopy. EDXS and EELS are methods which make us possible to quantitatively analyse the scanning-electron-microscopy or HRTEM images in terms of chemical composition. HRTEM is suitable for detailed grain boundary and multiphase analysis and is excellent for location of defects, foreign bodies and interfaces between two phases. In addiction with the focused ion beam (FIB) it is an excellent tool for a sample preparation on the atomic level. Lorentz microscopy can be used to image magnetic domains in a structure, whereas the 3DAP microscopy enables exact elemental positioning within a solid sample.
Laser confocal microscopy can be used to obtain both high resolution optical images (2D) and 3-dimensional profiles by capturing and layering multiple images through a range of depths within the sample via optical sectioning. This technique allows the user to acquire image data and images from complex 3D shapes including surface topology and profiling as well as internal structure imaging. Initial work on SPS compacts revealed small regions of cavitation throughout the sintered sample, leading to a density lower than the theoretical maximum (7.19 gcm-3 compared to 7.50 gcm-3). Using standard 2D laser imaging it was possible to locate one of these regions. The cavity was approximately 10 μm wide and 15 μm long. 3D laser microscopy on this area was performed. It could be observed that the centre of the cavity was very rough, consisting of a large pit of varying depth. It was also observed that there are multiple small holes/pits in the microstructure around the main cavity, which could be attributed to the redistribution of Nd-rich during the HDDR process from a large triple junction into a fine microstructure.
Task 6.3: Magnetic measurement (M3 - M35)
Magnetic measurements represented one of the most critical aspects of the project. We performed the so-called open loop measurements with the vibrating-sample magnetometer (VSM), and the closed loop measurement with the permeameter. One of the important tasks was to ensure that comparisons could be made between the two types of measurements; this is mainly because not every type of sample can be measured in both open- and closed- loop measurements. Powders, for example, could not be measured easily in a closed-loop permeameter measurement, whereas large samples, with dimensions greater than 5 mm (typically), could not be measured with an open-loop VSM.
A comparison between the results obtained by applying the two techniques was performed for various materials either utilized in sintered magnets (permeameter) or in a powder form (VSM) obtained after exposing the starting magnets to the HDDR process.
Task 6.4: Particle size and surface analyses (M6 - M30)
The SEM imaging was proven as very efficient for the assessment of the particle size, particularly for the powders, obtained by applying the wire explosion experiment, Although this technique does not make us possible to determine an exact particle-size distribution, it was evident that the main fraction of the particles had a diameter between 20 and 150 nm with occasional bigger particles of up to 2 μm, which surface can be analysed in detail. However, a more detailed analysis required the application of the laser diffraction, which should, in principle yield the respective distribution. But the method turned out not to be as useful as expected because of the agglomeration, partly ascribed to the magnetic attraction.
Task 6.5: Corrosion tests (M12 – M35)
Nd–Fe–B magnets require a protective coating / surface finish to minimize the effects of corrosion. Iron within the structure can ‘rust’, which causes a permanent structural change in Nd–Fe–B, which results in a permanent weakening of the magnetic performance – the worst case scenario is a total loss of magnetism. In order to determine the effect of corrosion on Nd–Fe–B magnets, numerous corrosion tests need to be employed. This is especially important for magnets that are supposed to be used in any type of electric vehicles.
For testing the corrosion resistance of magnets we had available multiple methods and tests. These were:
Autoclave: 120 °C / 2 bar / RH 100%
”HAST”: 130 °C / 2,7 bar / RH 95%
Environmental test: 85 °C / RH:85 %
Salt spray test (Coated magnets)
We had to apply the following test procedures:
Cleaning the samples
Weighing 1 + dimensions (surface area)
Test chamber (several days)
Weighing 2 => weight loss / area
The corrosion resistance can be improved by using various coatings or by reducing the oxidation rate of the Nd-rich phase in the case of the Nd-Fe-B magnets. The latter can be achieved by substituting the respective phase by some other phase, or by minimizing the electrochemical potential between this phase and the hard-magnetic phase Nd2Fe14B, which is the driving force for corrosion.
WP 7: Production of Permanent magnets and Thermal Management Components (M12 - M30)
WP Leader: Irena Škulj (MAGNETI), Jean-Marc Dubus (VALEO)
Summary of Results
Once we succeeded to prepare magnetic powders with required properties via the three routes covered in the work-packages WP3, WP4 and WP5, of which the HDDR+SPS route (WP5) turned out to be the most effective, we focused on the production of magnets. The work was mainly done in the magnet-producing company Magneti with the required equipment, experiences and knowledge. After considering various possibilities it was decided to apply the press-less sintering with the additional post-sintering heat treatment, which was proven as beneficial in order to further improve the properties. However, it was found that the results considerably depended on the processing parameters, hence they should be tuned for a particular processing set-up.
The thermal management was realized in Valeo, where it was decided to apply the conventional solution, based on an inverter, composed of the power module and the control card, making up possible to manipulate the machine-stator phases for the production of either the mechanical or electrical power.
There were no deviations from the objectives and task given in Annex I.
Details for each Task
Task 7.1: Preparation of magnet-processing (M12 - M20)
On the basis of the performed experiments we decided to include in the production process for the production of new magnets powder preparation (hydrogen decrepitation and milling) and heat treatments (sintering and heat treatment). Part of production for fine powders required HD plant and jet mill and magnet preparation part required moulds for powder packing and aligning and vacuum furnaces for sintering and heat treatments.
Requirements for equipment:
HD plant: For the furnace to be able to carry on Hydrogen decrepitation would be important to be able to reach temperatures up to 1000°C and pressures up to approximately 3 bar. HD plant should be able carry out HDDR process to be able to improve magnetic properties.
Jet mill: We tried to find the appropriate equipment for milling below 1 µm or less with Netzsch Company. The most promising proved to be CON JET. It was hard to define the needed specifications for the mill and milling medium.
Moulds: It was proved that moulds with cavities for NdFeB powders made of graphite were satisfactory. In the case of press-less process there is no need for a big press with a magnetic field. All that is needed is a graphite mould with cavities and magnetiser to align powders when already in the mould. With this kind of technology the production is limited to relatively simple shapes of magnets.
Furnace: Here we are talking about usual sintering furnace that is capable of reaching good vacuum and has options of fast cooling with Argon gas.
Task 7.2: Pressing, sintering and magnetization of samples (M14 – M24)
We decided for the press-less sintering technique. It was important to mill fine powders of the right chemical composition, and to sinter aligned powders into magnets. Throughout the process it was most important to follow the milling step by particle-size distribution analyser. Throughout the process we performed the tests of the following properties: chemical compositions, density, shrinkage, magnetic properties, microstructural changes.
Task 7.3: Assessment of permanent magnets (M20 – M26)
In order to determine the optimal processing conditions as well as processing techniques at particular steps of the procedure we compared different magnets. We found that the magnets sintered from ball milled powders and prepared through press-less method did not sinter at all; full density was not achieved. Therefore the magnets were not magnetic. For a comparison the powders were also pressed iso-statically and sintered. Magnets prepared via press-less method were too oxidized but magnet prepared via isostatic pressing had no hard magnetic phase Nd2Fe14B present in the microstructure. Powders prepared with ConJet in Netzsch were damaged during further processing and therefore not made magnets.
The final test was, of course, the magnetic measurement to prove the choice of the processing route and techniques. We compared the demagnetization curves of a magnet in as sintered state and after continued heat treatment. The measured properties were as expected:
Br = 1,21 T
bHc = 850 kA/m
iHc > 1600 kA/m
BHmax = 270 kJ/m3.
Task 7.4: Thermal management of power electronics (M18 – M28)
It turned out that the power electronics for controlling the electric-motor which consisted of the permanent magnets produced during the course of the MAG-DRIVE project was not as critical as it might be expected, therefore we sticked to the standard, water-cooled solution in terms of a commercially-available inverter. The inverter is composed of a power module which controls the machine-stator phases to produce mechanical or electrical power. The power module use depends on the voltage, either with the power Mosfet or IGBT that are driven by the control card.
The control card contains all the machine piloting algorithms in a microcontroller, as well as the rotor-position-sensor algorithms and the vehicle strategy, the battery state of charge, the machine temperature calculation, etc.
Task 7.5: Optimisation of EM power density using new permanent magnets (M24 - M30)
We selected the optimal sets of the magnets produced during the course of the project. As expected the best results were achieved by applying the HDDR+SPS route from the WP5. We were able to achieve the power densities of up to 2.56 kW/kG.
WP 8: Testing of the Magnets and Electric Motors in the Real-World Conditions (M24 - M35)
WP Leader: Jean-Marc Dubus (VALEO)
Summary of Results
This work-package was supposed to include the final research activities of the MagDrive project, since it was about testing and applying the magnets in electric motors, constructed in Valeo. It turned out to be completely successful since the delivered magnets passed all standard-prescribed tests, whereas the respective motor, performed according to specifications.
There were no deviations from the objectives and task given in Annex I.
Details for each Task
Task 8.1: Preparation for the tests (M24 – M26)
There were no special preparations needed for the tests of the magnets since all the required equipment was readily available at Valeo and Magneti. The magnetic properties were determined by applying the same techniques as described in the WP6, mainly by using the permeameter, designed for the so-called closed-loop measurements. More dedicated were the devices for the measurements of the bending force and displacement, torque, output efficiency, and thermal properties.
Task 8.2: Physical and magnetic properties of the permanent magnets (M26 – M28)
On the basis of the magnetic measurements on the powders, produced along the three routes, which were covered by the work-packages WP3, WP4 and WP5, we decided that the best magnets should result from the HDDR+SPS material. Hence we prepared bigger size magnets and cut them into smaller pieces that were appropriate for measurements and had their magnetic properties measured with permagraph at Magneti Ljubljana d.d.. Final magnets were additionally cut and ground to obtain a shape defined by Valeo, so that the diameters were about 40 mm. We ended up with 16 different magnets; four form each of the four different compacts related to the variation of the processing parameters. he results revealed a pronounced homogeneity of the compacts reflected in a reproducibility of the results. Whereas the magnets from all four compacts exhibited nearly the same remanence values of 0.8 T, the compact 1 was distinguished by the highest coercivity of 1100 kA/m, following by the compacts 3 and 4 with slightly lower values. The magnets from the compact 2 had about 10% lower coercivity, thus they were least applicable.
Task 8.3: Corrosion properties of the permanent magnets (M26 – M28)
The standard procedure in Valeo to check the corrosion resistance of electric components is the salt-spray test. It makes us possible to check the ring/brushes-compartment tightness and the vulnerability of the machine face to the electro-corrosion.
Test conditions:
Machine in rotation mounted in a specific angular position in a specific bench, with representative customer regulator plugs & an electrical circuit with a battery of 70 Ah, wiring of 35 mm2 2 m + 2 m,
Cycle: 75 min (60 min running on alternator mode (ON), 15 min stopped (OFF)) Phase « On »: Ambient temperature (Ta) Salt mass concentration Pulverization pressure Salted water quantity Rotation frequency (n) Alternator output (It) Alternator or system voltage (UB+)
Phase « Off »: Electric polarization Salt fog pulverization
: +35°C, : 5%, : 0.1 Mpa, : 1.5 ml/h, : 6 000 min-1,
: I /3 at 4 000 min-1/ +23°C, : Ureg,
: without, : without.
Cycles Number: 77 (equivalent of 96 h), For liquid cooled alternator, no liquid flow & electrical output defined in order to have Tj diode < Tj maxi.
The MagDrive magnets turned out to be suitable for the application, there was no magnet degradation, the test was passed successfully, and there were no performance losses.
Task 8.4: Tests on the new electric motor (M28 – M34)
The final tests were devoted to the evaluation of the performance of the electric-motor containing the MagDrive magnets. The measurements were carried out at different working temperature and rotation speeds. The most relevant output values were the torque and power, which were compared to the predictions of the simulations. The rotor equipped with the MagDrive magnets fulfilled the specifications. There was no demagnetization, the correlation between the simulation and measured results was good, hence all the objectives were achieved. The conclusion was that the MagDrive magnets, with reduced amount of heavy rare-earth elements, can be used in electric motors.
Potential Impact:
Dissemination and exploitation of project results are very important aspects of all European projects in order to present scientific and technological achievements to peers, financiers, industry as well as to the general public. The first step towards a better recognition of the intermediate and final results of any scientific activity is the dissemination on various levels. During the course of the MagDrive project the activities could be followed on the web page: http://mag-drive-fp7.eu. Besides a brief description of the project, its organization and the main objectives, it supplied the information about the current activities like the partners meetings and the workshops organized by the members of the consort, which were covered thematics related to the scope of the project. The workshops were open to the public and often attended by researchers or Ph.D. Students, particularly from the host institutions or companies. Those events served their purpose by spreading the information about the MagDrive project and about the EC funding scheme in general. In order to present the project in a more popular way, maybe targeting a younger generation of the future scientists, economists as well as end users, we established also a related Facebook page, where we published not just about the MagDrive but also general news from the field of magnetic materials, electric vehicles and similar topics. Of course, the scientific outcome of the project was presented in a more common way too. So far, we have published two peer-reviewed papers, given four invited talks, given six talks on scientific conferences, and presented four posters on scientific conferences. This sort of dissemination is still an ongoing activity and the list of publications is not closed yet since some of the papers are still under review, and some presentations are still to be given. The MagDrive activities were informally advertised during the the so-called doors-open days at the involved institutions. For example, at the Jozef Stefan Institute, where the project was coordinated, the doors-open days take place in March every year, attracting several thousands visitors of all ages from nursery children to seniors. It is of vital importance to inform the public about the research in the frame of the projects like MagDrive, which have a direct impact on environment (electric vehicles) and economy (lowering of the production costs- heavy-rare-earth-free magnets).
While the benefits of dissemination might be regarded as an investment for the future, the exploitation is expected to have an almost immediate impact. On the basis of the fruitful collaboration during the course of the MagDrive project, Valeo and Kochanek Entwicklungsgesellschaft initiated a new industrial project that is planed to be profitable for both partners. The improvements of the spark-plasma-sintering technique, which were developed and tested for the purposes of the MagDrive project, the patent applications “Flash Spark Plasma Sintering of fully dense, nanostructured permanent magnetic materials” was prepared by Queen Mary, University of London and University of Birmingham. This method, in a combination with the hydrogen-decrepitation-deabsorption-recombination (HDDR) process, was found to be most upscalable among the considered routes for a massive production of permanent magnets. Hence we prepared a respective detail exploitation plan, where we first compare the lab-scale and industrial-level production.
Lab-scale production
The lab-scale production of large FSPS NdFeB compacts is carried out along the following route:
1. Manual preparation of graphite die with raw material, taking 5 minutes
2. Loading die into SPS, pumping to vacuum and performing SPS on the prepared die to make pre- compact. Cooling & removal of die, taking 12 minutes
3. Manual removal of pre-compact from die and removal of graphite foil from compact, taking 5 minutes
4. Manual loading of new tooling and pre-compact into the SPS, pumping SPS to vacuum and heating to pre-heat temperature, taking 5 minutes
5. Pre-heating of compact, Flash-Sintering and cooling, taking 6.2 minutes
6. Manual removal of compact from SPS and removal of graphite foil from compact, taking 5 minutes
The overall required time is 38.2 minutes. At this rate it should be possible to produce 12 compacts during an 8 hour shift. Each compact can then be cut to produce 4 magnets. During the Mag-Drive project, the cutting of the compacts to specified dimensions was performed by external contractors. The rate of production of finished magnets from the compacts was roughly one magnet per hour. Hence, at lab scale using existing tooling, magnet production by FSPS takes approximately 1hr 40min per magnet.
Based on this, it is clear to see that the rate of production is limited by numerous factors, including: All manual preparation of dies and unloading/cleaning of sintered compacts. Pumping the furnace to vacuum. Heating the furnace to the sintering or pre-heat temperature.
Waiting for the die/compact to cool before the next die/compact can be loaded. Cutting and finishing of magnets from the compacts.
In addition, the cutting operation leads to the loss of around 50 % of the compact as scrap; since rectangular magnets are being cut from a circular disk of material and since the edges of the compacts need to be avoided due to cracking and increased porosity in these areas. The cracking and porosity at the edges of the compacts is due to the increased loss of heat from these regions during FSPS. Graphite felt is placed around the edges of the compact to reduce this effect but due to the lack of control over the temperature of the felt, this does not eliminate the problem all together.
Scaling up to industrial level
Based on the above assessment of lab scale production: a continuous, automated process to produce net-shape forged magnets would be the most cost-effective and resource-efficient method to work towards in the scale-up to industrial level. Lab scale FSPS experiments with Nd-Fe-B have shown that very large and rapid degrees of deformation can be achieved using only small applied pressures. This highlights the potential for efficient net-shape forging by FSPS; provided that the die used is electrically insulated from the FSPS tooling and is heated. This subsequently avoids the uneven heating and surface cracking problems associated with the lab scale process. The proposed processing route consists of four steps:
1. Filling of pre-press die by automatic hopper system, taking two minutes
2. Cold pressing, taking eight minutes
3. IR ‘biscuit’ firing & preheating, taking 60 minutes
4. Net-shape FSPS, taking 10 minutes
The estimated total time is 80 minutes without incorporating cooling into the “per-magnet” processing time, since multiple magnets can be left to cool at the same time; hence this is not a rate-limiting step.
The continuous aspect of the process ensures a high throughput of material. Since a vacuum is maintained at all times and there is no waiting for the chamber to reach the pre-heat temperature, the time and energy for processing is cut considerably. Automation minimises the time and expense from manual labour operations. Using cold compaction and biscuit firing instead of a pre-SPS step to produce the green compact reduces the pre-compact fabrication time by around 21.5 min; particularly due to the fact that die preparation and the removal of graphite foil from the sample are no longer necessary. Finally, the net shape forging step removes the need for post-fabrication machining, which significantly reduces the magnet fabrication time and eliminates raw material waste all together. As shown above, the total time for the fabrication of each magnet using this process, is 80 seconds. This would equate to the production of 360 magnets per day, based on one 8 hr shift.
Of note is that net-shape forming also provides added value to the finished magnets, particularly if more complex shapes could be formed. Currently, complex-shape magnets are made from ferrite materials which have been machined into the final shape. The machining process results in a lot of material loss and is therefore only economically viable when applied to cheap materials. If net- shape materials could be produced without any material wastage, then it would become economically viable to produce complex-shape Nd-Fe-B magnets. This would be advantageous in e.g. electric motors; where the volume of magnetic material required to produce the same power as the equivalent ferrite-based machine would be roughly halved. Smaller, lighter and more powerful machines would then be possible.
Continuous, multi-chamber, ‘tunnel type’ SPS machines are already in existence, manufactured, for example, by NJS Co. Ltd.
The figure of 360 magnets per 8 hr day is based upon a single line of production; with only one magnet being Flash Sintered at any one time. There is no reason why the SPS machine and tooling could not be developed so that several magnets could be simultaneously Flash Sintered (using e.g. multi-cavity cold pressing and FSPS tooling systems), multiplying the machine output. 6-10 cavity dies have already been employed in regular SPS. Additionally, back-to-back shift work could be introduced to double the daily output. With this in mind, the machine and tooling could be developed to produce a throughput of anything between 360 and 7200 magnets per day.
As demonstrated, the upscaling of the processing procedure is feasible. In the following we focus on the cost analysis, which is divided into the cost of the industrial FSPS, and into the cost of the raw material.
Cost analysis of industrial scale FSPS
In consultation with Kennametal UK Ltd., who own and run a single-chamber industrial scale SPS, costs have been drawn up for the purchase and running of an industrial scale continuous SPS.
The cost of a new, customized, multi-chamber SPS as discussed above is roughly EUR 1.74M. For the FSPS process suggested, an additional chamber for the hopper loading and cold pressing of the Nd-Fe-B powder would have to be developed and added to the machine; at an estimated maximum cost of EUR 580k. The investment required for the research and development of the custom machine would be roughly EUR 232k. With attendant costs of building, services and other incidental one-off expenses, the total one-off cost for the customized SPS machine and installation is expected to be around EUR 2.9M:
Machine research and development investment: EUR 232k
New multi-chamber SPS furnace and associated equipment, in accordance with FSPS requirements and desired throughput: EUR 2.74M
Building / Services to accommodate new SPS facility: EUR 580k
Purchase of handling aids to help in periodic removal of large numbers of magnets from the machine and to refill the hopper with powder: EUR 232k
Ongoing running costs associated with the SPS include:
Part of the time of a trained technician to periodically load the hopper with powder and remove cooled magnets from the machine. Considering two technicians working back-to-back for overnight production, this could equate to a pro rata cost of EUR 35k per annum.
Replacement of custom made FSPS graphite tooling and purchase of different sets for different desired magnet shapes. For the proposed net shape FSPS process this only relates to the punches being used. Given the relatively low pressures required for the small magnets the punches are expected to last for up to 200 runs. It is therefore recommended that the punches be changed twice a day. Considering the highest output case of the multi-processing of 10 magnets at once, this equates to 100 punch sets a week and 5200 punch sets in a year. The cost for one set of custom designed punches is EUR 58, hence this is an ongoing expense of EUR 302k per year. Potential savings could be made by producing the punch sets in-house.
Ongoing SPS maintenance and repairs, EUR 23-35k per year.
Energy bills for running of the SPS machine. The FSPS process consumes around 0.02 kWh per magnet, which at a maximum throughput of 7200 magnets per day, would equate to 144 kWh per day or 37,440 kWh per annum. Factoring in other associated energy costs (e.g. the intermittent running of a vacuum pump and chiller), the annual energy consumption is expected to be roughly 70,000 kWh/annum; costing around EUR 8k:
Technician time: estimated EUR 34.7k per year
Graphite tooling: estimated EUR 300.5k per year
SPS maintenance and repairs: estimated EUR 23-34.7k per year
Energy bills: estimated EUR 8.1k per year
which sum up to the estimated total cost of maximum EUR 377.9k per year.
Raw materials cost
The current cost to make and ship a standard 12g Nd-Fe-B magnet containing 10 wt.% Dy is around EUR 1.2. This price is based on analysis of the current Chinese local market and cost when supplied to Europe, as determined in consultation with Valeo. Of this figure, the raw materials cost accounts for EUR 1.056.
For a Dy-free composition, as per the current MQU-F material being processed by FSPS, the raw materials cost would be EUR 0.4 per magnet. For a maximum output of 7200 magnets per day, this equates to an annual raw materials cost of around EUR 760,000.
The above assumptions lead us to the final estimate.
Production cost per magnet and & break-even point
Taking into account the ongoing and raw materials costs, and neglecting the initial one-off costs, the cost of production per magnet equals EUR 0.608. If the magnets were retailed at a competitive market price of EUR 1.20 each plus VAT, this would equate to a profit of EUR 0.592 per magnet and EUR 1,108,224 per annum.
The break-even point, at which the initial one-off investment costs equal the profit, would therefore be expected to occur at the end of the third quarter of the third year after starting the above-described industrial production.
In summary, We demonstrated that the transfer of the spark-plasma sintering (SPS) method for the production of permanent magnets without heavy-rare earth content, which was developed during the course of the MagDrive project, could be commercially successful, which is the single most exploitation achievement of the project.
An eventual single company, based on the above-described process, would offer several hundreds new working positions within EU with a possibility for further growth, depending on the market demands. These will become big, if the vision of electric vehicles dominating the roads, as promoted by the MagDrive project, comes true.
List of Websites:
http://mag-drive-fp7.eu
Electro-motors may replace internal-combustion engines in the future vehicles. It is therefore of a vital importance to invest in development of electric drive. The working principle of any electric motor is based on a magnetic-torque-induced rotation, hence it is indispensable to provide a magnetic field, which can be produced either by coils or by permanent magnets. The formers make us possible to continuously control the field up to extremely high values by varying the applied electric current, whereas the magnets operate without any external power and can be shaped almost arbitrarily to fit into any motor design. Modern high-performance permanent magnets are made of alloys, which contain rare-earth elements. Due to the economical and political situation in the world the availability of these materials is limited. Particularly severe is the situation in the case of the so-called heavy-rare-earth elements, like dysprosium and terbium, which are added to the basic materials to enhance the performance of the magnets up to the level defined by the specifications of the electric motor. In order to reduce the manufacturing costs and even to prevent unexpected delays in production, it is necessary to invest in development of corresponding materials with a reduced amount of heavy-rare-earth elements, which is the main objective of the MAG-DRIVE project.
Further requirements for the material are related to the specific working conditions and environment: the temperatures of about 100 0C and corrosive atmosphere, in which the electric motors installed in vehicles operate. The next objective is therefore to make a magnet not just without using heavy-rare-earth elements and with the specified performance, but also with an ability to perform at such conditions. And the final goal is to integrate the magnets into an electric motor and/or an alternator with an improved and specially developed thermal management and power electronics, suitable as a power aggregate for electric vehicles.
The performance of a magnet is characterized by its hysteresis loop, which represents the response on an external magnetic field. The area surrounded by the loop is related to the energy stored in an operating magnet, hence it should be as large as possible. Consequently the value of the external field, required to demagnetize the magnet, called coercivity, and the output of the magnet in the absence of the external field, called the remanence have to be high enough. Additionally, the squareness of the hysteresis loop impacts the surrounded area too. Although these parameters depend on the intrinsic properties of the starting material, they can be substantially influenced by the processing of the powders and the way how the powders are consolidated to form solid magnets. Whereas the remanence can be controlled by the density and alignment of the compacts, the standard way to provide a high coercivity used to be based on adding heavy rare-earth elements to the starting powder materials. An alternative is to keep the size of the grains within the particles as small as possible. There are several ways to achieve this goal. The most obvious seems to provide the powders with sufficiently small particles. In this manner we approached along two different routes, either by means of the so-called wire-explosion procedure, where the fine powder is formed upon applying a high-electric-current pulse, or using various sophisticated techniques. Since the chemical composition of the powder obtained from the wire explosion does not match the on the hard-magnetic phase it was necessary to compactify the powders in terms of the Kochanek process, which made us possible to modify the composition by keeping the particle size low. The powders resulting from milling were sintered in a standard way after carefully tuning the respective temperatures, heating rates and durations. Another option to deal with the small grains is in the frame of the so-called HDDR process, which stands for the hydrogen decrepitation, deabsorption, recombination. Basically, it is a method where hydrogen gas induces a decay of the material, which then reforms with an improved morphology. The so-obtained powders were finally exposed to the spark-plasma sintering (SPS) to make the solid magnets.
Within the MagDrive project we investigated the three processing routes, which at the end all led to to the demonstrator magnets with desired properties. However, it turned out that the magnets obtained from the HDDR+SPS route had the best characteristics, therefore the demonstrator electric-motor was built with them. The motor passed all required tests and could be in principle applicable in real vehicles.
The results in the investigations within the MagDrive projects were presented in several peer-reviewed publications and on some major scientific conferences. We applied for one patent related to the details of processing. In order to demonstrate that the developed procedure is not important just from the scientific point of view, we proved its economic importance by preparing a financial plan for the exploitation of the HDDR+SPS procedure for a massive production of magnets. It is expected that the break-even point, at which he initial one-off investment costs equal the profit, would occur at the end of the third quarter of the third year since the start of the production.
Project Context and Objectives:
The MAG-DRIVE consortium consisted of research partners: Jozef Stefan Institute (Slovenia), University of Birmingham (United Kingdom), Institute for Chemistry, Technology and Metallurgy (Serbia), Queen Mary University of London (United Kingdom), of a research and development company (SME) Kochanek Entwicklungsgesellschaft (Germany), of a magnet-producing company Magneti Ljubljana d.d. (Slovenia), and of a producer of components for automotive industry Valeo (France). Jozef Stefan Institute (JSI) is the leading Slovenian research institute with a lot of experiences in production and characterization of magnetic materials by means of magnetic measurements and electron microscopy. JSI was the coordinator of the project, contributed to all activities with an emphasize on the magnetic measurements and electron microscopy. The University of Birmingham is ranked in the top 100 universities worldwide. The Magnetic Materials Group in the School of Metallurgy and Materials has a reputation as one of the leading groups in the field, focused on development of novel processing techniques, which was also their main role in the project, combined with magnetic measurements and material characterization. The University of Belgrade's Institute of Chemistry, Technology and Metallurgy (ICTM) is leading institution in the field of micro-system and nano-system technologies in the Western Balkan. They mainly performed mechanical milling and characterization of powders. Queen Mary, University of London is one of the UK's leading research focused higher education institutions. The Functional Nanomaterials Group has particular expertise in processing including the spark-plasma sintering, which was their contribution to the project. Kochanek Entwicklungsgeselschaft is a small privately held SME, specialized in radical new concepts for manufacturing nano-crystalline metal parts, some of them were applied in the project. Magneti Ljubljana d.d. is privately owned SME for making permanent magnets. They provided conventional sintering and some magnetic measurements. VALEO is an independent industrial group fully focused on the design, production and sale of components, integrated systems and modules for the automotive industry. They prepared and performed all sorts of tests of the demonstrator magnets, and built the electro-motor with the required power electronics.
The description of work was divided into nine work-packages (WPs).
The WP1: Project Management lasted through the whole course of the project, and it was about the project management, and it was completely covered by the Jozef Stefan Institute. The main goals were to:
Ensure efficient project coordination, progress monitoring, budget and financial control
Ensure the quality of all activities and the timely delivery of reports and deliverables
Ensure effective and on-time communication between the project consortium and the European Commission through the REA
Prepare the rolling 6-monthly Action Plan
The activities were divided into the following tasks:
Task 1.1: Project management (M1 – M36)
Task 1.2: Preparation of 6-monthly Action Plan (M1 – M30)
Task 1.3: Creation and maintenance of project website (M1 – M36)
The results, entirely contributed by the Jozef Stefan Institute, were presented in 3 deliverables.
The WP2: Technical requirement lasted from the month 1 to the month 8, and it was led by Valeo. It was devoted to a detailed survey of industrial specifications and technical requirements for the demonstrator and the new electric motor, as described in the following tasks:
Task 2.1: Benchmarking electric motors (M1 – M3)
Task 2.2: Design of thermal management for power electronics (M3 – M8)
Task 2.3: Assessment of currently available permanent-magnet (M1 - M8)
The results, contributed by all partners, were presented in 4 deliverables.
The development of material was covered in the work-packages WP3, WP4 and WP5.
The WP3: Route I-The oxide route was led by Kochanek Entwicklungsgesellschaft, and it lasted from the month 1 to the month 30. The primary objective was to produce nano-crystalline hard-magnetic material by applying the wire-explosion experiment and the Kochanek process. The investigations comprised the following tasks:
Task 3.1: Preparation of test cores from rare-earth oxides (M1 – M12)
Task 3.2: Preparation of hard magnetic materials (M9 – M18)
Task 3.3: Preparation of complex 3D-shaped permanent magnets for industrial use in Evs (M15 – M30)
Task 3.4: Development of theory for hard magnets based on the Kochanek process (M18 - M30)
The results, contributed by Kochanek Entwicklungsgesellschaft, Jozef Stefan Institute and University of Birmingham, were presented in 6 deliverables.
The WP4: Route II-The nanoparticle route was led by the Institute for Chemistry, Technology and Metallurgy, and it lasted from the month 1 to the month 30. The main goal was to obtain magnetic powder with the particle size of the order of magnitude of 1μm by means of innovative milling techniques through the following tasks:
Task 4.1: Analysis of the nano-milling technologies (M1 - M6)
Task 4.2: “Light-gas” milling and comparison studies (M6 – M20)
Task 4.3: Separation of magnetic particles (M12 – M24)
Task 4.4: Production of Dy-diffused Nd–Fe–B magnets (M20 – M30)
The results, contributed by the Institute for Chemistry, Technology and Metallurgy, Jozef Stefan Institute, Valeo and University of Birmingham, were presented in 5 deliverables.
The WP5: Route III – The HDDR+SPS route lasted from the month 1 to the month 30, and it was led by the University of Birmingham. The goal was to increase the grain size by applying the hydrogen-decrepitation-deabsorpition-recombination (HDDR) process in order to achieve high coercivity, and to spark-plasma sintering (SPS) to make the magnets as described by the following tasks:
Task 5.1: HDDR processing conditions (M3 – M15)
Task 5.2: Alignment and pressing of green compacts (M9 – M18)
Task 5.3: Investigation of densification kinetics during SPS (M12 – M24)
Task 5.4: Spark-plasma sintering of aligned green compacts (M18 – M30)
The results, contributed by University of Birmingham, Jozef Stefan Institute, the Institute for Chemistry, Technology and Metallurgy and the Queen Mary, University of London, were presented in 6 deliverables.
The WP6: Microstructural and magnetic characterisation lasted from the month 3 to the month 35, and led by the Jozef Stefan Institute, provided the service for WP3, WP4 and WP5 with constantly ongoing activities by means of analysing the structural and magnetic properties of the processed materials. The tasks were:
Task 6.1: Scanning electron microscopy (M3 – M35)
Task 6.2: Advanced electron microscopy (M3 – M35)
Task 6.3: Magnetic measurement (M3 – M35)
Task 6.4: Particle size and surface analyses (M6 – M30)
Task 6.5: Corrosion tests (M12 – M35)
The results, contributed by Jozef Stefan Institute, Valeo, Kochanek Entwicklungsgesellschaf, University of Birmingham, the Institute for Chemistry, Technology and Metallurgy and the Queen Mary, University of London, were presented in 18 deliverables.
The WP7- Production of permanent magnets and Thermal Management Components was led by Magneti and Valeo, and it lasted from the month 12 to the month 30. The aim was to make the magnets from the powders prepared in the processing WP's, and to construct the power electronics for controlling the electric motor. The activities were described in the following tasks:
Task 7.1: Preparation of magnet-processing (M12 – M20)
Task 7.2: Pressing, sintering and magnetization of samples (M14 – M24)
Task 7.3: Assessment of permanent magnets (M20 – M26)
Task 7.4: Thermal management of power electronics (M18 – M28)
Task 7.5: Optimisation of EM power density using new permanent magnets (M24 – M30)
The results, contributed by all partners, were presented in 5 deliverables.
The WP8: Testing of the Magnets and the Electric Motors in the Real-World Conditions, led by Valeo, and lasted from the month 24 to the month 35, represented the final research work package, since the output was a working electric motor, based on the MagDrive magnets. The work was divided into the tasks:
Task 8.1: Preparation for the tests (M24 – M26)
Task 8.2: Physical and magnetic properties of the permanent magnets (M26 – M28)
Task 8.3: Corrosion properties of the permanent magnets (M26 – M28)
Task 8.4: Tests on the new electric motor (M28 – M34)
The results, contributed by Valeo, Jozef Stefan Institute and Magneti, were presented in 5 deliverables.
The WP9: Dissemination and exploitation, was led by the Jozef Stefan Institute and lasted through the entire course of the project. It was devoted to the dissemination and exploitation of the MagDrive results in various forms according to the proposed tasks:
Task 9.1: Dissemination (M1 – M36)
Task 9.2: Exploitation (M1 – M36)
The results, contributed by all partners, were presented in 3 deliverables.
The objectives of the project were valid during the entire duration of the project with differing intensity.
A: The focus of the overall objectives was targeted towards:
developing high-coercivity hard-magnetic materials without heavy-rare-earth elements like dysprosium or terbium by dramatically reducing the grain size to below 1 μm using three novel processing routes
using the developed materials to produce demonstrator magnets applicable for electric-vehicle motors
optimisation and the thermal management of the power electronics and rotors in electric motors
efficient project coordination, quality of all activities, timely delivery of reports and deliverables, effective and on-time communication between the project consortium and European commission, preparation of rolling 6-monthly Action plan
B: Objectives related to short-term activities and progress:
to provide industrial specifications and the technical requirements for the magnet demonstrator and the new electric motor, which was completely achieved in the WP2
to build up nanocrystalline hard-magnetic materials in principles of the wire-explosion experiment, which was completely achieved in the WP3.
to provide magnetic powder of particle size in the order of magnitude of 1 μm by means of “light-gas” milling procedures, which was completely achieved in the WP4.
to produce HDDR-processed Nd-Fe-B anisotropic magnetic powders with the grain sizes in the sub-micron region, which was completely achieved in the WP5.
to constantly provide a characterisation of structural and magnetic properties of the materials, produced in WP's 3,4 and 5 by means of electron microscopy and state-of-the-art magnetic measurements, which was completely achieved in the WP6.
to prepare the magnet-processing equipment, and to perform the pressing, sintering and magnetization of samples, which was completely achieved in the WP7.
to define and to report on the testing procedures, and to report on the structural, magnetic and corrosion-resistance tests, which was completely achieved in the WP 8.
to maintain external project communication, and to guarantee the impact of the project through the creation of a comprehensive exploitation plan, which was completely achieved in the WP9.
The flow-chart of the project included four different milestones:
MS1: Technical Specification for New Magnets defined within the WP2, led by Valeo was expected to be achieved by the end of the month 8
MS2: Midterm Review defined within the WP1, led by the JSI was expected to be achieved by the end of the month 18
MS3: First Demonstrator Assembled defined within the WP7 led by Magneti was expected to be achieved by the end of the month 23
MS4: Project Finished defined within all WP's led by the JSI was expected to be achieved by the end of the month 36
Project Results:
The S&T results were obtained in the work-packages WP2, WP3, WP4, WP5, WP6, WP7 and WP8. Within the WP2 we set the technical requirements for the demonstrators. The WP3, WP4 and WP5 represented three different routes towards the same goal to produce magnetic powders with the specified properties. The powders were consolidated in the form of solid magnets within the WP7, which was also devoted to the construction of the power electronics, required to control the electric motor, build with the magnets within the WP9. The WP6 was horizontal, and it covered all magnetic and structural characterization, required in other research-and-development-oriented WP's.
WP 2: Technical Requirements (M1 - M8)
WP Leader: Jean-Marc Dubus (VALEO)
Summary of Results
All activities within the WP 2 were finished in the first eight months. According to the plan we performed a detailed survey of industrial specifications and technical requirements for the demonstrator and the new electric motor. It was found that there were two competitive technologies in terms of performance and price, both with some advantages and disadvantages. Available products from the leading manufacturers were analysed in detail. The problem of cooling was addressed not just by designing but already by providing some prototypes for the required electronic circuits. Finally, we obtained the information about the currently available permanent magnets to be build in the electric motors.
There were no deviations from the objectives and tasks given in Annex I.
Details for each Task
Task 2.1: Benchmarking electric motors (M1 – M3)
Electrical vehicles can be configured in a number of different ways. Based on various considerations, for example, required output power and the available space. A choice was made between integrated starter generators (ISGs) and belt-drive starter generators (BSGs):
The main advantage of the ISG machine is that it can be used for ZEV driving and to maximize the regenerative braking power, as the thermal engine could be disconnect through a clutch mechanism.The main drawback of the ISG machine is the mechanical integration in between the combustion engine and the gearbox, which increases the operating costs.The comparative analysis of the competitor machines focused on the following: power density, efficiency, high torque at low speed in E-drive, cooling system, use of raw materials (in particular the use of RE), machine technology
We tested machines from the following companies:
Honda
Porsche
Mercedes
Cadillac
Lexus
Toyota
Kia
These motors where purchased on the open market from the OEMs. We tested some electrical motors that fall into the following categories:
Mild hybrid electrical vehicle (HEV)
Mild & Full HEV
Full Hybrid with Plug-in
The motors were tested as received and then stripped and cut away for imaging and other tests on the materials in the motors. Detailed tests were carried out on motors from:
Porsche (manufactured by Bosch)
Toyota
Renault (manufactured by Continental)
VW (manufactured by ZF Sachs)
The motors ranged in terms of the power output from 20 kW to 45 kW. Sizes were in the range 200 to 300 mm. Special attention was given to the position and orientation of the magnets. The Toyota motors all used embedded magnets with a “V” configuration. The German motors favoured a surface positioning of the magnets. It is important to note that the Continental motor, produced for the Renault Kangoo contains no magnets. Of course this motor has a large weight penalty as a result. The best power density was achieved by the Toyota Yaris electrical machine. This is mainly due to the use of high-grade Nd-Fe-B magnets and a high stator-filling ratio of 2,5kW/kg of raw material. The lowest power density was measured for the Renault Kangoo motor due to the wound rotor: 1kW/kg.
The main advantages of belt-driven machines are that they fit in the place of standard alternator, they are cheap and could achieved 85% of the CO2 reduction at about half of the price of an ISG system. However, the main drawback is that they are not as efficient as the ISG architecture in regenerative-braking mode. In the comparative analysis of our competitors we focused on the following: power density, efficiency, high torque at low speed in E-drive, cooling system, use of raw materials (in particular rare earths), machine technology.
In the MAG-DRIVE BSG investigation we looked four belt-driven devices. The electrical machines were produced by Hitachi, Bosch and Mobis. These motors where purchased on the open market from the OEMs. The rated power ranged from a low 5 kW to 82 kW, 32 kg Bosch SMG 180. These machines were all about 20cm long.
Task 2.2: Design of thermal management for power electronics (M3 – M8)
Under the consideration were two types of the water-cooled inverters which can operate at the temperatures up to 80 0C: the 24 V B2-9 related to the crankshaft motor generator and the 270 V charger-inverter type CMG B2-6. Both Devices were developed by VALEO.
Crankshaft Motor Generator B2-9:
Voltage range 36/55 Vdc
Typical voltage 48Vdc
Maximum phase AC current 350Arms
Temperature range -40°C / 80°C Water cooled
270V Charger inverter specification CMG B2-6
Typical voltage 270Vdc
Voltage range 220/300 V dc
Maximum phase AC current 260Arms
Temperature range -40°C / 80°C Water cooled
Task 2.3: Assessment of currently available permanent-magnet (M1 - M8)
RE-Fe-B-based magnets were taken from some of the motors and sent for a full chemical analysis. We noticed a remarkably high Dy content in excess of 10 wt.%. This is approximately twice as high as we were expecting. The quantities of the other elements were in line with expectations.
WP 3: Route I – The Oxide Route (M1 - M30)
WP Leader: Wolfgang Kochanek (KE)
Summary of Results
The primary objective of the WP 3 was to produce nanocrystalline hard-magnetic material by applying the Kochanek process which is based on the reduction of oxides. The results of extensive experimental investigations revealed that samarium oxide could not be reduced in any form at any conditions, hence the conclusion was drawn that the samarium-cobalt magnets could not been made by means of the oxygen-reduction process. It was also demonstrated that the reduction of neodymium oxide was relatively low, therefore an additional infiltration of the neodymium nanoparticles is required in order to get the optimum results. It was decided to achieve this goal by performing the so-called wire-explosion experiments, where the required nanoparticles are formed upon the applying an extremely high electrical voltage between the two ends of a ribbon made of respective material. The resulting pulse current yields a decay of the ribbon into nanoparticles. The required ribbons were produced by spark-plasma sintering. Finally, we applied the Kochanek process to prepare the demonstrator magnets. The analysis, based on the scanning-electron microscopy and magnetic measurements, proved the route, developed in the frame of the WP3, promising for a massive production of permanent magnets, although further improvements in a reduction of the oxygen content, enhancement of the density, and of the amount of the hard NdFeB phase in the magnets are still required.
There were no deviations from the objectives and task given in Annex I.
Details for each Task
Task 3.1: Preparation of test cores from rare-earth oxides (M1 - M12)
The Kochanek process is based on the reduction of metal oxides. Therefore the feed stock for the Sm2Co17 consists of Sm2O3 and Co3O4 with organic compounds and of the Nd2O3, Fe3O4 and B with organic compounds in the case of the Nd2Fe14B. The main challenge consists in the reduction of the rare-earth oxides which have a high thermodynamic stability. The Gibbs free energy are 1085,3 kJ/mol and 1009,6 kJ/mol at room temperature for samarium and neodymium, respectively, indicating that the reduction under this conditions is thermodynamically not favourable With rising temperature the Gibbs free energies drop only a little. At 1800 K they still reach 844,5 kJ/mol for samarium and 858,6 kJ/mol for neodymium which does not significantly improve the thermodynamically conditions for reduction and suggests the equilibrium to be far on the left side of the reaction.
In the case of samarium a series of experiments was conducted which produced a total of over 40 samarium oxide based samples. The researched parameters in these reductions include temperature in a range from 650°C up to 1360°C, reduction duration from 3 h up to 6 h, precursor density from 3.29 g/cm3 up to 4,19 g/cm3 and different compositions of additives. Regardless of all the variations of parameters the weight loss measurements clearly proved that no samarium oxide reduction could be achieved as demonstated by means of the XRD spectroscopy. Hence, we conclude that no samarium oxide can be reduced in order to apply the Kochanek process for making the Sm2Co17 magnets.
In the case of neodymium a series of experiments was conducted which produced a total of over 250 neodymium oxide based samples. The researched parameters in these reductions include temperature in a range from 650°C up to 1360°C, reduction duration from 3 h up to 9 h, precursor density from 1,47 g/cm3 up to 3,49 g/cm3, precursor particle size from roughly around 100 nm up to 5 μm, homogeneity of precursor material varied through the preparation process and different compositions of additives. Depending on these parameters a neodymium oxide turnover rate from 0% up to 20% could be achieved. The turnover rate was determined through the weight loss of the samples during reduction process. Simple by hand prepared samples achieved if at all only a low turnover rate. The highest neodymium oxide reduction rate could be reached through application of the more complex Kochanek process which yielded smaller particles and a higher homogeneity. At high temperature and long duration these samples reached a turnover rate close to 20%. The sample consisted of Nd2O3, Fe3O4, B and organic additives and was prepared according to the Kochanek process. The starting density was 3.11g/cm3 with a height of 6.06 mm and a diameter of 13.00 mm. After reductive treatment at 800 °C for 4 h a complete iron oxide turnover was achieved but none of the neodymium oxide. The density was lowered to 2.45 g/cm3with a height of 6.17 mm an a diameter of 12.63 mm. After pressing with 4.5 t/cm2 the sample showed a density of 4.47 g/cm3 with a height of 3.15mm and a diameter of 13.00 mm. A secondary reduction at 1360°C for 5 h yielded a neodymium oxide turnover rate of 8.6% determined through weight loss and a final density of 6.21 g/cm3 with a height of 2.86 mm and a diameter of 11.5 mm. Compared to the theoretical density of 7.65 g/cm3 the material reached 81.2%. The XRD analysis of the produced samples reflect the results obtained through weight loss measurements. Even though XRD does not allow to determine an exact neodymium oxide turnover rate the found phases depend on the composition and thus on the turnover rate. Samples treated with lower temperature whose weight loss measurements show no or little neodymium oxide turnover rate, show only iron and neodymium oxide phases in their corresponding XRDs. XRD results for materials prepared along the Kochanek process with high reduction temperature and a neodymium oxide turnover rate close to 20% showed beside a main neodymium oxide and iron phase a small pure Nd phase. The EDS analysis showed a significant higher neodymium oxide turnover rate for oxides close to the surface. It was found that iron rich areas as well as iron lean areas achieved a similar neodymium vs. oxide ratio of about 7 : 3. This corresponds to a neodymium oxide reduction rate of 71% for oxides close to surface. The magnetic analysis of the samples show according to the previous results only a soft magnetic response due to the high iron content. The samples showed no coercivity and therefore no hard magnetic phases have formed.
We conclude that the reduction of neodymium oxide in the presents of Fe/Fe3O4 could be achieved up to 20% indicating that the thermodynamic stability is different in the presence of iron compared to pure neodymium oxide. To achieve a competitive hard magnetic NdFeB-material a neodymium oxide turnover rate close to 100% is needed since small impurities in the material have a negative influence on the magnetic properties. Trying to push the equilibrium of the reduction by high temperature is limited as we found out that the matrix starts to melt beyond1400°C. Which is supported by the iron neodymium phase diagram. Therefore we proposed an additional infiltration of the neodymium nanoparticles by performing the so-called wire-explosion experiments. The nano-sized powder was formed upon applying an extremely high electrical voltage between the two ends of a ribbon made of respective material. The resulting pulse current yielded a decay of the ribbon into nanoparticles. The required ribbons were produced by melt spinning.
Task 3.2: Preparation of hard magnetic materials (M9 – M18)
We used super-structured co-precipitated Nd2O3 /Fe3O4 material and conducted a series of experiments which produced a total of over 60 super-structured co-precipitated samples. The experiments where conducted in a temperature range from 700°C up to 1300°C and a duration of reduction of 4h. Samples chemical reduced with a temperature lower than 900°C re-oxidized on air at room temperature. The lower the reduction temperature the more rigorous the samples re-oxidized. This well known effect is based on the lesser sintering at lower temperatures resulting in a more reactive part. It is noticeable for the co-precipitated materials that the re-oxidation occurs at reduction temperatures up to 900°C. Comparable samples based on mere oxide mixtures did not re-oxidise at the researched temperatures with the lowest temperature being 600°C. This shows a more fine-grained particle size distribution for the co-precipitated materials. A SEM-imaging comparison between a neodymium oxide powder and the co-precipitated material illustrates the difference in particle size and therefore the reactivity due to a higher surface area. While the neodymium oxide particles size was above one micron the particle size of the co-precipitated material was within the sub-micron range. Regardless of all the temperature variations and the higher reactivity based on smaller particles the weight loss measurement results of all super-structured co-precipitated precursors showed a significant lower turnover rate instead of an expected increase in turnover. These losses in turnover rate can be explained through the XRD results. Instead of forming a Nd2Fe14B phase the reduction led to a Fe main phase, a unexpected NdFeO3 phase, a Nd2O2S phase based on some sulphate impurities and a small pure Nd phase. Due to the formation of a NdFeO3 phase only an incomplete iron oxide reduction was achieved. This explains the significant lower turnover rate compared to samples based on oxide mixtures which in general achieve a close to 100% iron oxide reduction.
Task 3.3: Preparation of complex 3D-shaped permanent magnets for industrial use in Evs (M15 – M30)
During a wire explosion the “wire” reaches a temperature of more than 104 K with a heating rate of around 107 K/s. In general the explosion is achieved by the discharge of a capacitor through the conductive wire. Their capacity ranges from a fraction of a μF up to several hundred μF with storages voltages varying from a few kV to several hundred kV. Current peaks up to 50kA are common. We monitored the voltage and current time dependence for a wire electrical explosion of a steel wire done with the experimental setup built for the MAG-DRIVE project. The capacitor was charged with 6kV and while discharging the current reached a maximum peak of almost 8kA. The whole process took less than 300μs.The explosions took place in a special reactor, build for the purposes of the MAG-DRIVE project. The explosion chamber was affixed inside a steel frame. Screwed on the whole outside of the steel frame were 3 mm thick steel plates to shield against electro magnetic waves. The explosion chamber was rendered inert with an constant Argon stream. Also hooked to the chamber was a pump that circulates the contained gas. In between the circulating gas stream was a washing flask to collect the product. The size of the resulting particles was determined by means of the scanning-electron microscopy yielding the assessment that the main fraction had a diameter between 20 and 150 nm with an occasional deviations up to 2 μm. The low density iron part for the infiltration with Nd based nanoparticles was produced according to the Kochanek process. The starting density of the disc shaped iron Kochanek parts was in the range of 2.72 to 2.98 g/cm3, which equals a free pore volume of 61.4% to 58.8%. For infiltration a well stirred suspension of stabilized neodymium based nanoparticles was added drop-wise onto the Kochanek parts. To minimize the exposure to air the procedure was done in an argon ventilated glove box. The dropping speed of the suspension was adjusted to be as fast as the infiltration into the Kochanek sample itself. Dropping was carried out for as long as the suspension was sucked in.
The concentration of nanoparticles in the suspension needs to be 27,87% (v/v) for a free volume of 61.4% and 31.06% (v/v) for a free volume of 58.8% to be infiltrated in one session. Infiltration tests showed that suspension with such high concentrations of particles have a poor infiltration behavior. The parts will not get infiltrated thoroughly. Because of that a nanoparticle concentration of 15% (v/v) was used, which showed a good infiltration behavior up to the centre of the part in the preliminary tests.
The use of low concentration slurries made it necessary to infiltrate repeatedly. In between infiltration sessions the parts were dried until the cyclohexane evaporated. The procedure was repeated until the calculated amount of suspension was infiltrated into the Kochanek part.
After infiltration the samples were heat treated to take the organic components out of the part on the one hand and on the other hand to form the NdFeB hard magnetic phase. To evaporate the amines and to decompose the remaining organic components the samples were heat treated between 400 °C and 500 °C for several hours under hydrogen. Afterwards they were heated up to 850 °C for a duration of 10 hours under nitrogen to form the NdFeB hard magnetic phase. The parts reached a density from 3.85 g/cm3 up to 4.26 g/cm3.
Task 3.4: Development of theory for hard magnets based on the Kochanek process (M18 - M30)
During the course of the project it was established that the reduction of oxygen might not be the most efficient procedure for the production of magnetic powders containing rare-earth and transition-metal elements. Therefore we successfully applied an equally innovative alternative by means of the wire-explosion experiment. However, we were still able to explain the reasons for a failure of the initially-proposed process in terms of a plausible theory.
The idea was to reduce neodymim oxide by exposing the samples to the elevated temperatures in the range from 650°C up to 1400°C in the H2, N2 and/or Ar atmosphere. At temperatures above 1000°C the reduction of neodymium oxide starts to begin. Unexpectedly at reduction temperatures of 1200°C and higher the turnover rates caves in not even reaching a complete iron oxide reduction anymore. This decrease in turnover rate was accompanied with a melting of the sample surface. The molten surface could also be the reason for the low turnover rate, since it could be responsible for inhibiting the reduction by blocking the gas transfer between the centre of the sample and the surrounding atmosphere. An EDS analysis of the sample surface visualized the molten surface as dark grey areas that mainly consisted of iron. Even though the XRD and EDS results verified that a neodymium oxide reduction could be achieved to some degree there was no hint of a Nd2Fe14B hard magnetic phase in these results. Magnetic measurements of the samples confirmed these previous results as only a soft magnetic response, due to the high iron content, could be measured.
WP 4: Route II: The Nanoparticle route (M1 - M30)
WP Leader: Dana Vasiljević-Radović (ICMT)
Summary of Progress towards Objectives
The main objective of the WP4 was to obtain magnetic powder with the particle size of the order of magnitude of 1μm by means of innovative milling techniques. Towards this aim we analysed the commercially available equipment and decided for the most suitable technology which would serve as the starting point for further improvements. The application of this technology gave the satisfactory results in terms of the particle size close to the micrometer region. We quantitative analysed the size distribution and separated the powder.
There were no deviations from the objectives and task given in Annex I.
Details for each Task
Task 4.1: Analysis of the nano-milling technologies (M1 - M6)
The state of the art of different milling techniques that can be used to obtain sub-micrometer-sized nano-crystalline rare earth-transition metals magnetic materials was surveyed. Nanocrystalline NdFeB magnetic materials were considered since they possess the highest energy product of all the permanent magnets available today. We reviewed properties of different mechanical milling methods, including impact milling, ball milling, jet milling, roller milling, etc. We put the largest stress to the ball media milling and jet milling. The consideration includes the basic phenomena occurring in different milling procedures, the available equipment, including pros and cons for each type, different accompanying effects that may influence the quality of milling and the state of the art of experimental works in the field. Basic properties of NdFeB sub-micrometer and nano-powders were reviewed from the point of view of their influence on the production of permanent magnets. We performed also numerical simulation of different grinding and milling procedures by applying the the discrete element method. Based on the survey including the properties of both the powder material to be processed (brittleness, flammability, toxicity, etc) and the available techniques, we decided for the equipment most suitable for the use in our own experiments with NdFeB. It is a FRITSCH Planetary Micro Mill PULVERISETTE 7 fabricated by Fritch GmbH from Idar-Oberstein, Germany. The reasons for the decision are the following:
The grinding of materials down to the nano-meter range requires very high application of energy and therefore significantly higher rotational speeds than those allowed by typical planetary mills. Conventional planetary ball mills are characterized by grinding bowls that are clamped to the sun disk. This limits the maximum possible rotation speed because of a specific speed limit, the centrifugal forces acting on the bowls will be so great that the clamping of the bowls can no longer hold. Sinking of the grinding bowls into the sun disk of the mill solves these problems. In this construction, the centre of gravity of the bowls lies in the plane of the sun disk. The centrifugal force arising generate significantly lower overturning moment, which in turn allows the mill to run at significantly higher speed. As a result, this mill attains a speed of up to 1100 rpm, reaches centrifugal accelerations of up to 95 times the force of gravity, making the energy application approximately 150 % greater than conventional planetary mills. This significantly reduces the grinding time to reach the nano-meter range. For certain materials, as is NdFeB, brittle and hard, we think that only this level of energy application even allows the milling to NPs range. Grinding bowl size is 80 ml, ant it is made from tungsten carbide (WC). Grinding balls are from the same material. There are 2 bowls, max. sample quantity is 5ml and min. sample quantity is 0.5 ml. In bowls dry and wet grinding can be performed. We bought special emptying device which enables a quick and easy separation of the grinding balls and suspension. Grinding bowl is tightly sealed so that even grinding in suspension without any additional sealing is possible. Lead is equipped with pressure and temperature sensors and has a bleeder valve so that any overpressure in the bowl can be equalized in a controlled fashion. These features allow grinding in an inert atmosphere.
Task 4.2: “Light-gas” milling and comparison studies (M6 – M20)
The milling experiments were performed by the use of a JET-mill. The milling was accomplished by the use of light-gas milling of HD Nd-Fe-B material using a production-scale JET-mill. The comparative experiments were performed in the company Netzsch from Hanau in Germany, where they applied their ConJET dry grinding machine.
The starting material for the JET milling experiment was chosen to be a Magneti-standard NdFeB material with a low Dy content (1 wt. %). This material was chosen due to low intrinsic coercivity (HcI)- owing to the low Dy content. 50 kg of NdFeB flakes was HD processed for the experiment,
The smallest ConJet produced by Netzsch, ConJet 10, was used for the experiments. In contrast to a fluidized-bed Jet mill, ConJet does not have a fluidized milling chamber with a fluidisation nozzle on the bottom and a classifier on the top. The milling chamber is located around the classifier, with an increased number of nozzles. The milled powder particles path is therefore set longitudinal to the classifier, by which a sharper cut point of larger particles can be achieved compared to a fluidized-bed JET mill. In order to avoid oxidation of the incoming HD flakes and the collected (milled) powder, the whole mill was enclosed in a flexi glass bell, filled with inert atmosphere. The milling parameters were as follow:
• Milling gas N2, 99.95% purity (open circuit).
• Nozzle diameter: 1.2 mm.
• Nr. of nozzles: 5.
• Nozzle pressure 7 / 8 bar
• Classifier speed: 16,000 RPM/ 12,000 RPM / 3,000 RPM / 6,000 RPM / 9,000 RPM.
• Milling chamber pressure: 0.2–0.29 bar
The following two experiments were declared as a successful run:
• Ex1: Classifier: 16,000 RPM / p(nozzle): 8 bar / productivity: 0.2 kg/ h / D(observed) < 3 μm
• Ex2: Classifier: 6000 RPM / p(nozzle): 8 bar / productivity: 2 kg/ h / D(observed) < 8 μm
Task 4.3: Separation of magnetic particles (M12 – M24)
The first step in the magnetic-particles separation is the analysis of the powder size. We performed the on-site experiments with the SEI LOT Phenom PRO. The Ex1 (6,000 RPM / p(nozzle): 8 bar / productivity: 0.2 kg/ h) and the Ex2 ( 6000 RPM / p(nozzle): 8 bar / productivity: 2 kg/ h) materials were analysed. Although the Ex2 material obviously consisted of small particles, the productivity was too low, hence the average measured particle size was between 3 and 5 μm, which is very close to the submicron region, estimated as the optimum to achieve the desired magnetic properties.
Task 4.4: Production of Dy-diffused Nd–Fe–B magnets (M20 – M30)
The essential step was to separate the particles of the desired size from larger suspension volumes using a process combining sedimentation and membrane filtration. NdFeB particle suspension after surfactant-assisted wet milling is allowed to sediment for time t, then the upper portion of the suspension corresponding to height h is decanted and filtered in protective argon atmosphere under pressure ΔP through a membrane with a specific pore size. The process is repeated n times with the settled slurry and filtrate returning to another milling cycle. The filtration cake is composed of particles having a size distribution determined by sedimentation and filtration parameters. After filtration, polycarbonate membrane with the filtration cake consisting of NdFeB particles was immersed in 25 mL of chloroform to dissolve the membrane and disperse the particles. The resulting suspension was sonicated for 5 min and investigated with Zeta Sizer Nano ZS90 using a 1:10 dilution ratio. The resulting particles were characterized by the average diameter of 315 nm with most particles falling in the range from 200 to 500 nm. Such magnetic powder exhibits the desired properties, the only obstacle for an eventual massive production are relatively small quantities of the produced material, which calls for further investigations towards scaling up the involved procedures.
WP 5: Route III: The HDDR+SPS route (M1 - M30)
WP Leader: Allan Walton (UOB)
Summary of Results
Hydrogenation Disproportionation Desorption and Recombination (HDDR) is a high temperature hydrogen treatment that utilizes the ability of materials such as Nd-Fe-B to readily absorb and desorb hydrogen at elevated temperatures which results in a fine grained material with very interesting magnetic properties.
It was found that the optimal processing conditions for the material with a reasonably low Dy content were the temperature of about 880 °C, with a disproportionation pressure of 1500 mbar, recombination under vacuum and quick cooling upon completion. However, by increasing the Dy as well as the the Co content the required applied pressure should be increased in order to complete disproportionation.
In order to fully exploit the potential of the material, it is highly desirable to align the particles prior to the final compaction. The result of the conventional spark-plasma sintering is an isotropic green compact, therefore we developed a method to pre-align the starting HDDR powder in the presence of an external magnetic field as described in detail within the Task 5.2.
There were no deviations from the objectives and task given in Annex I.
Details for each Task
Task 5.1: HDDR processing conditions (M3 - M15)
In this task the processing conditions for each step of the HDDR process were optmized for a range of different Nd-Fe-B magnet compositions including primary cast materials and secondary scrap sintered materials. The cast material includes a range of ternary alloys that can be altered with alloying additions to optimize the HDDR reaction. A range of sintered magnets have been used with low Dy levels.
The aim of this task was to avoid over-processing of the Nd-Fe-B material during each step of the process without foregoing the anisotropy that can be developed in HDDR powders. The disproportionation step was monitored using mass-flow controllers and a thermocouple for each chosen composition. The impact of the processing temperature and hydrogen pressure has been assessed in terms of their ultimate value, ramp rate and hold time during the disproportionation and recombination reactions. The impact of each of these variables on the microstructure has been determined with the aim of minimizing the size of the Fe, Fe2B and the Nd hydride rods in particular, prior to desorption. In a similar fashion the pressure-reduction rate and quench rate for the desorption step was monitored in order to minimize grain growth on formation of the Nd2Fe14B phase during the recombination reaction. The magnetic properties of the HDDR powders were measured in WP6 to assess the level of anisotropy that was developed for each composition and set of processing parameters.
Three compositions of Nd-Fe-B magnets were chosen for the trials. Two sources from large sintered blocks of Composition A - Nd13.4Dy0.8Al0.7Nb0.3Fe78.5B6.3 (at %) and Composition B – Nd12.5Dy1.8Al0.9Nb0.6Co5.0Fe72.8B6.4 (at %). The first composition was very low in Dy as the target of this project is to lower the Dy content of fully dense sintered magnets. The second composition was chosen as the Dy content is slightly higher but still below the average Dy content measured across a range of mined sources (Wang 2013). Following on from this '′real′′ scrap magnets were processed based upon voice coil motor magnets from hard disk drives. These were chosen as hard disk drives represent the largest source of sintered Nd-Fe-B from consumer products, they are collected in large quantities for data removal and there is relatively little variation in composition from product to product.
The HDDR processing conditions have been optimized for the processing of 20g sintered Nd-Fe-B batches of Composition A (low Dy). This was achieved by changing the process temperature (830-930 °C), disproportionation pressure (1000-2000 mbar), recombination pressure (0-350 mbar) and recombination time (0-30 mins) in order to produce the greatest magnetic properties. The optimal process conditions for Composition A were found to be at 880 °C, with a disproportionation pressure of 1500 mbar, recombination under vacuum and quick cooling upon completion. The voice coil motor magnets with a similar composition also processed successfully within the optimized parameters. Composition B which contains higher Dy and Co content, however, required a higher hydrogen pressure in order to complete disproportionation. By increasing the disproportionation pressure to 2000 mbar as required by Composition B, samples of Composition A exhibited lower remanence and were less anisotropic due to over-processing.
Having tracked the disproportionation reaction of the various sintered, book mould and strip cast starting alloys, it could be observed that the reaction kinetics were strongly affected by composition, rare earth content and initial microstructure. It was found that for the cast starting materials, the microstructure had the largest effect on the disproportionation reaction when compared to rare earth content. Strip cast alloys initiate disproportionation at lower hydrogen pressures than conventional book mould cast alloys which could be attributed to the much smaller grains in the strip cast material allowing for fast propagation of the disproportionation reaction through the matrix grains. Increasing neodymium content in ternary alloys increases the rate of reaction for hydrogen decrepitation but delays the onset of disproportionation. This can be attributed to the excess of Nd-rich grain boundary phase which reacts readily during hydrogen decrepitation but acts as a transport medium for hydrogen before disproportionation initiates within the matrix grains. For sintered magnets the additions of Co and Dy also delay the onset of the disproportionation, requiring an increased hydrogen pressure to initiate the reaction.
Task 5.2: Alignment and pressing of green compacts (M9 - M18)
For the production of aligned green compacts suitable for spark-plasma sintering (SPS) compaction, anisotropic HDDR powder was produced. The HDDR powder was a mixture of coarse and fine particles when removed from the processing furnace, and so it required breaking down using a pestle and mortar to remove any agglomeration. After 5-10 minutes of light grinding, the powder was suitable for green-compact production. 15 g of the HDDR powder was then loaded into a neoprene isostatic press bag and filled to tap density. A rubber bung was then pushed into the top of the press bag to a tight fit and then sealed using electrical tape. The seal was to ensure that liquid could not be introduced to the sample during the pressing stage and also prevents oxidation when the sample is prepared under an inert atmosphere. The sealed press bag was then placed inside the coil set from a capacitor-discharge pulse magnetiser. This piece of equipment consists of a large bank of capacitors that when pulsed from a fully charged state produce a 9 T field through the copper coils in the direction from ceiling to floor. The sample was situated centrally within the coil and held in place using a wooden-support system to ensure that the press bag is kept completely vertical. If the press bag was to lean to the side or be allowed to move then the alignment of the powder within the press bag would not be perfect. The capacitors were charged and discharged three times to pulse the powder particles three times. This allowed for full alignment of the powder particles in the upwards direction. The press bag containing the aligned powder sample was then placed inside the water-filled chamber of the isostatic press. The top ram was then inserted into the chamber and the pressure increased to 10 tons and held for 1 minute. The pressure was then very slowly released to prevent cracking of the sample as the walls of the neoprene isostatic press bag released from the surface of the sample. Once removed from the isostatic press chamber, the green compacts were ejected from the press bag and inspected. The green compact does not have a parallel flat top and bottom faces. This is due to a combination of the particle size of the starting powder, the magnetic field exhibited by the compact and the lack of a binding agent as used in bonded magnets. The powder particles in these regions follow the magnetic field lines of the bulk of the magnet, this reduces the density of the upper and lower regions of the green compact. There are a number of ways in which this problem could be addressed. Mechanical removal of the ‘fluffy’ end parts using a blade is possible; however, the overall alignment of the sample could be reduced during the cutting process as the powder particles are disturbed. The large variation in particle size could be impacting on the stacking of the particles during compression. By reducing the particle size it could be possible to produce better compaction under isostatic pressing.
When reducing particle size it was important to consider an appropriate milling technique. Wet milling techniques such as low-energy roller ball milling requires cyclohexane carrier fluid and milling times from 1-20 hours to produce a fine particle size. The powder would then become very fine, leading to an increased potential for oxidation and yield loss during the drying process in a vacuum port. Jet milling would produce a very fine powder with particle sizes close to 1 μm and below. These particles would be very susceptible to oxidation and unfortunately due to the low sample size (20g) jet milling is out of the scope of this work.
The milling process chosen for this work was a laboratory sized hammer mill. This milling technique utilises three rotating blade arms that grind the powder against a ring of teeth that line the internal wall of the mill. A 0.2mm hole size mesh is placed in the bottom of the mill for the milled powder to pass through and into a collection tube. The built-in speed control unit allows for careful control of the milling process to avoid sparking of the Nd–Fe–B during the grinding of the particles between the rotating arms and the integrated grinding teeth.
The hammer mill was run for 10 minutes to allow all of the HDDR powder to pass through the 0.2 mm mesh. Due to the depth of the teeth in the hammer mill, the HDDR powder became caught during the milling process, reducing the yield collected in the tube; however, blowing compressed air through the loading hopper released the powder from the teeth to continue the milling process. The milled HDDR powder was then passed through the isostatic pressing route described above. The reduced particle size allowed for better stacking of the powder and hence a more uniform green compact was produced. There was a small amount of fine particles remaining in the bottom of the press bag after the green compact was removed, however they did not form along the magnetic field lines, as in the previous samples. However, due to the lack of a binding agent and the nature of the pressing route the powder around the edge walls of the surface became loose as the compact was handled; therefore, an alternative pressing technique was also considered.
Uniaxial pressing using a 20-mm die set allows for a higher pressure to be used, as well as controlled specimen dimensions due to the polished internal wall of the die set and the upper and lower punch pieces. As with the conventional SPS compaction route, this preparation method produces isotropic samples as there is no current alignment field. However, the pressing force used with this press and die set is 20 tons, rather than the 0.5 ton compaction used with the graphite die sets. It can be observed that the green compact is very uniform in size and shape; however, as previously stated the magnet is isotropic in nature.
A possible solution to this is that a copper wire could be wound around the die set and an electrical current be passed through the wire during the pressing action to create a magnetic alignment field within the die set. Unlike the isostatic alignment route, where the powder is pulsed aligned before pressing, this route would maintain a constant alignment field throughout the entire pressing process; however, the alignment field would be much weaker.
A potential improvement for the isostatic pressing route would be to powder blend the HDDR powder with milled neodymium hydride powder. The hydride powder would serve two roles in this process, firstly due to the powder being soft and friable, it would potentially acts as a binding agent to keep the green compact intact during and after removal from the isostatic pressing bag. Also it would aid with liquid-phase sintering during the SPS process, improving the densification behaviour of the compact.
Task 5.3: Investigation of densification kinetics during SPS (M12 – M24)
The initial experiments were performed on large batches (400 g) of the low-Dy material with the composition Nd13.4Dy0.8Al0.7Nb0.3Fe78.5B6.3 since the main target of the project is to reduce the rare-earth content in the fully-dense sintered magnets. The magnetic properties, including anisotropy were reduced compared to the 20 g batches due to the highly endothermic nature of the recombination reaction. The temperature of the sample can drop by up to 100 °C during recombination, hence taking the sample out of the optimal processing window, leading to a loss in anisotropy. The grain size was submicron however; hence it was a good material for initial SPS testing i.e. subsequent grain growth during SPS treatment could be checked.
The green compacts for the SPS treatment were formed by lightly grinding the HDDR-treated sample with a pestle and mortar. The powder was then loaded into a neoprene isostatic press bag and sealed with a rubber bung and electrical tape. The bag was then placed into a capacitor discharge pulse magnetiser and pulsed three times with a 9T alignment field. The sample was then pressed to 10 tons using an isostatic press. The green compact was then removed from the press bag and wrapped with a single layer of graphite paper and pressed into a graphite die. The die was then loaded into the SPS machine which was sealed and evacuated.
A range of SPS treatment conditions were trialled, and the best results were found after processing at 900 °C for 1 minute, the sample was cooled under vacuum and ejected from the graphite die. The density of this sample was calculated as 7.14 g cm-3, which was measured with a densitometer. The theoretical maximum density for Nd-Fe- B is ~7.5 g cm-3, so the sample was ~95% dense. The microstructure of the samples were investigated by means of the SEM technique. We observed a mixture of very fine grains along with some large grains throughout the material. The average grain size was measured using the Feret diameter, and was calculated to be 0.75 μm with a standard deviation of 0.41 μm.
Task 5.4: Spark-plasma sintering of aligned green compacts (M18 - M30)
Like the conventional sintering, the spark-plasma sintering (SPS) is used to bind particles of material together into strong useable objects. This occurs through mass transfer at the atomic scale, acting to build ‘necks’ between particles. In addition to heating, pressure and electric fields are present during SPS to achieve the goal. The relatively high temperatures and sintering durations required to fully consolidate powders using conventional sintering techniques render the process unsuitable for the sintering of certain materials. For some materials, the sintering temperatures necessary to achieve full density are too high to sinter using conventional methods. In others, unwanted chemical reactions may occur during heating for long periods of time; and, where it is desirable for the properties or certain characteristics of the starting powder to be transferred to the final sintered compact, quicker processing times and/or lower processing temperatures are necessary to avoid any change in microstructure. For this reason, much of modern advanced materials sintering involves the use of Field Asisted Sintering Techniques (FAST). In contrast to the slow external heating methods employed in conventional sintering, FAST techniques employ the use of an applied electric current to produce high heating rates of up to 2000 K/min. Such heating rates are achieved through the application of pulsed current (from a few micro seconds to a few milli seconds) of extremely high current density - up to several kA. This is possible due to the high electrical conductivity of the tooling involved and is therefore achieved without the need for high voltages (typically below 10 V). Powders and green compacts are usually sintered within a conductive die. In the case of an electrically conductive green body, energy is dissipated directly within the sample and the conductive die; whilst for electrically insulating green parts, the heat is generated through the Joule heating of the die and is transmitted by conduction to the powder. The process yields a densification (shrinkage) of the exposed material. We applied the heating rate of 100 °C min-1 up to 900 °C and pressure of 50 MPa for four different powders.The fall of the rate of densification to 0 beyond 730 °C suggests that densification has completed. After sintering, the sample density was measured using Archimedes principle and confirmed to be near theoretical, hence applicable for the production of magnets.
WP 6: Microstructural and magnetic characterisation (M3 - M35)
WP Leader: Matej Komelj (JSI)
Summary of Results
This work package provided the service for WP3, WP4 and WP5 with constantly ongoing activities by means of analysing the structural and magnetic properties of the processed materials. In term of structural characterization the basic tool was the scanning electron microscope which makes us possible to qualitatively observe the microstructure, i.e. the phase composition, resulting from different processing conditions. By applying this technique we demonstrated the strength of the HDDR method in reducing the amount of the undesired soft-magnetic alpha-Fe phase. Advanced microscopy methods made us possible to quantitatively characterize the chemical compositions of the considered materials. Other techniques, like high-resolution transmission electron-microscopy (HRTEM) or laser confocal microscopy were used for imaging on atomistic level or of complex 3D shapes, which was important for the spark-plasma sintered compacts as presented in the description of the Task 6.2. Of equal importance as the microscopy were magnetic measurements. We relied on two different principles: the open-looped vibrating-sample magnetometer and the closed-loop permeameter. The former was more suitable for the samples in a powder form, whereas the latter was applicable for solid magnets. Both techniques were widely used throughout during the execution of the project. The particle size and surface morphology were of a great importance for the WP4. Different milling conditions lead to different particle sizes. It was very important to protect the magnets for electrical-vehicle applications against corrosion. An appropriate protection can drastically reduce a decay of a magnet due to environmental conditions as demonstrated in the Task 6.5.
There were no deviations from the objectives and task given in Annex I.
Details for each Task
Task 6.1: Scanning electron microscopy (M3 – M35)
Scanning electron microscopy is the single-most important analysing technique for materials science . The length scales involved at every stage of the procedure, from hundreds of microns to hundreds of nanometres, are well covered by this technique. The key information that was supplied here by the SEM technique relates to grain size and phase distribution. Of particular importance were the observations relating to book-mould-cast and strip-cast material. The SEM revealed the striking differences in microstructures that can be obtained with these techniques. Book-mould casting tended to produce larger grains, a coarser microstructure and more separated phases – and most importantly the presence of free iron, which is potentially at least very problematic for permanent-magnet production of any kind. The SEM studies revealed how strip casting, to a large extent, overcomes these problems.
By way of benchmarking the SEM technique was also applied to sintered magnets, where it was able to show that sintered Nd-Fe-B magnets also consist of two main distinguishable phases: Nd2Fe14B matrix grains surrounded by a network of Nd-rich triple junctions and grain boundaries. When SEM is used to look at HDDR material, which is material on a sub-micrometre scale it shows very nicely that the HDDR process is successful in ‘removing’ the α-Fe from the starting microstructure. When the HDDR process is applied to strip cast material the SEM is able to show that the long, columnar grains are also absent from the microstructure and that this HDDR processed strip cast Nd-Fe-B could lead to better compaction using spark plasma sintering (SPS) as the regions of cavitation are not as significant as in HDDR processed book mould cast Nd-Fe-B. The SEM was also used to look at SPS samples of HDDR powder. This starting powder was found to have an ideal submicron grain size distribution, typical of HDDR-processed material. However, there was also evidence of explosive grain growth in some of these samples, particularly those processed at high temperatures. Again, SEM is an ideal technique for assessing materials in this way and provides a clear explanation for the lower-than-expected coercivity values, but with the advantage of providing us with a clear path to overcome the problem. In the case of the best samples achieved so far for HDDR powder and SPS, based on grain size alone, the retention of a proportion of fine grains which are of single-domain size in the best samples suggests that the magnetic properties of these samples can be improved.
Task 6.2: Advanced electron microscopy (M3 - M35)
Advanced electron microscopy includes techniques like energy-dispersive X-ray spectroscopy (EDXS), electron-energy loss spectroscopy (EELS), high-resolution transmission electron microscopy (HRTEM), Lorenz microscopy, Laser assisted 3D atom probe (3DAP) microscopy, and Laser confocal microscopy. EDXS and EELS are methods which make us possible to quantitatively analyse the scanning-electron-microscopy or HRTEM images in terms of chemical composition. HRTEM is suitable for detailed grain boundary and multiphase analysis and is excellent for location of defects, foreign bodies and interfaces between two phases. In addiction with the focused ion beam (FIB) it is an excellent tool for a sample preparation on the atomic level. Lorentz microscopy can be used to image magnetic domains in a structure, whereas the 3DAP microscopy enables exact elemental positioning within a solid sample.
Laser confocal microscopy can be used to obtain both high resolution optical images (2D) and 3-dimensional profiles by capturing and layering multiple images through a range of depths within the sample via optical sectioning. This technique allows the user to acquire image data and images from complex 3D shapes including surface topology and profiling as well as internal structure imaging. Initial work on SPS compacts revealed small regions of cavitation throughout the sintered sample, leading to a density lower than the theoretical maximum (7.19 gcm-3 compared to 7.50 gcm-3). Using standard 2D laser imaging it was possible to locate one of these regions. The cavity was approximately 10 μm wide and 15 μm long. 3D laser microscopy on this area was performed. It could be observed that the centre of the cavity was very rough, consisting of a large pit of varying depth. It was also observed that there are multiple small holes/pits in the microstructure around the main cavity, which could be attributed to the redistribution of Nd-rich during the HDDR process from a large triple junction into a fine microstructure.
Task 6.3: Magnetic measurement (M3 - M35)
Magnetic measurements represented one of the most critical aspects of the project. We performed the so-called open loop measurements with the vibrating-sample magnetometer (VSM), and the closed loop measurement with the permeameter. One of the important tasks was to ensure that comparisons could be made between the two types of measurements; this is mainly because not every type of sample can be measured in both open- and closed- loop measurements. Powders, for example, could not be measured easily in a closed-loop permeameter measurement, whereas large samples, with dimensions greater than 5 mm (typically), could not be measured with an open-loop VSM.
A comparison between the results obtained by applying the two techniques was performed for various materials either utilized in sintered magnets (permeameter) or in a powder form (VSM) obtained after exposing the starting magnets to the HDDR process.
Task 6.4: Particle size and surface analyses (M6 - M30)
The SEM imaging was proven as very efficient for the assessment of the particle size, particularly for the powders, obtained by applying the wire explosion experiment, Although this technique does not make us possible to determine an exact particle-size distribution, it was evident that the main fraction of the particles had a diameter between 20 and 150 nm with occasional bigger particles of up to 2 μm, which surface can be analysed in detail. However, a more detailed analysis required the application of the laser diffraction, which should, in principle yield the respective distribution. But the method turned out not to be as useful as expected because of the agglomeration, partly ascribed to the magnetic attraction.
Task 6.5: Corrosion tests (M12 – M35)
Nd–Fe–B magnets require a protective coating / surface finish to minimize the effects of corrosion. Iron within the structure can ‘rust’, which causes a permanent structural change in Nd–Fe–B, which results in a permanent weakening of the magnetic performance – the worst case scenario is a total loss of magnetism. In order to determine the effect of corrosion on Nd–Fe–B magnets, numerous corrosion tests need to be employed. This is especially important for magnets that are supposed to be used in any type of electric vehicles.
For testing the corrosion resistance of magnets we had available multiple methods and tests. These were:
Autoclave: 120 °C / 2 bar / RH 100%
”HAST”: 130 °C / 2,7 bar / RH 95%
Environmental test: 85 °C / RH:85 %
Salt spray test (Coated magnets)
We had to apply the following test procedures:
Cleaning the samples
Weighing 1 + dimensions (surface area)
Test chamber (several days)
Weighing 2 => weight loss / area
The corrosion resistance can be improved by using various coatings or by reducing the oxidation rate of the Nd-rich phase in the case of the Nd-Fe-B magnets. The latter can be achieved by substituting the respective phase by some other phase, or by minimizing the electrochemical potential between this phase and the hard-magnetic phase Nd2Fe14B, which is the driving force for corrosion.
WP 7: Production of Permanent magnets and Thermal Management Components (M12 - M30)
WP Leader: Irena Škulj (MAGNETI), Jean-Marc Dubus (VALEO)
Summary of Results
Once we succeeded to prepare magnetic powders with required properties via the three routes covered in the work-packages WP3, WP4 and WP5, of which the HDDR+SPS route (WP5) turned out to be the most effective, we focused on the production of magnets. The work was mainly done in the magnet-producing company Magneti with the required equipment, experiences and knowledge. After considering various possibilities it was decided to apply the press-less sintering with the additional post-sintering heat treatment, which was proven as beneficial in order to further improve the properties. However, it was found that the results considerably depended on the processing parameters, hence they should be tuned for a particular processing set-up.
The thermal management was realized in Valeo, where it was decided to apply the conventional solution, based on an inverter, composed of the power module and the control card, making up possible to manipulate the machine-stator phases for the production of either the mechanical or electrical power.
There were no deviations from the objectives and task given in Annex I.
Details for each Task
Task 7.1: Preparation of magnet-processing (M12 - M20)
On the basis of the performed experiments we decided to include in the production process for the production of new magnets powder preparation (hydrogen decrepitation and milling) and heat treatments (sintering and heat treatment). Part of production for fine powders required HD plant and jet mill and magnet preparation part required moulds for powder packing and aligning and vacuum furnaces for sintering and heat treatments.
Requirements for equipment:
HD plant: For the furnace to be able to carry on Hydrogen decrepitation would be important to be able to reach temperatures up to 1000°C and pressures up to approximately 3 bar. HD plant should be able carry out HDDR process to be able to improve magnetic properties.
Jet mill: We tried to find the appropriate equipment for milling below 1 µm or less with Netzsch Company. The most promising proved to be CON JET. It was hard to define the needed specifications for the mill and milling medium.
Moulds: It was proved that moulds with cavities for NdFeB powders made of graphite were satisfactory. In the case of press-less process there is no need for a big press with a magnetic field. All that is needed is a graphite mould with cavities and magnetiser to align powders when already in the mould. With this kind of technology the production is limited to relatively simple shapes of magnets.
Furnace: Here we are talking about usual sintering furnace that is capable of reaching good vacuum and has options of fast cooling with Argon gas.
Task 7.2: Pressing, sintering and magnetization of samples (M14 – M24)
We decided for the press-less sintering technique. It was important to mill fine powders of the right chemical composition, and to sinter aligned powders into magnets. Throughout the process it was most important to follow the milling step by particle-size distribution analyser. Throughout the process we performed the tests of the following properties: chemical compositions, density, shrinkage, magnetic properties, microstructural changes.
Task 7.3: Assessment of permanent magnets (M20 – M26)
In order to determine the optimal processing conditions as well as processing techniques at particular steps of the procedure we compared different magnets. We found that the magnets sintered from ball milled powders and prepared through press-less method did not sinter at all; full density was not achieved. Therefore the magnets were not magnetic. For a comparison the powders were also pressed iso-statically and sintered. Magnets prepared via press-less method were too oxidized but magnet prepared via isostatic pressing had no hard magnetic phase Nd2Fe14B present in the microstructure. Powders prepared with ConJet in Netzsch were damaged during further processing and therefore not made magnets.
The final test was, of course, the magnetic measurement to prove the choice of the processing route and techniques. We compared the demagnetization curves of a magnet in as sintered state and after continued heat treatment. The measured properties were as expected:
Br = 1,21 T
bHc = 850 kA/m
iHc > 1600 kA/m
BHmax = 270 kJ/m3.
Task 7.4: Thermal management of power electronics (M18 – M28)
It turned out that the power electronics for controlling the electric-motor which consisted of the permanent magnets produced during the course of the MAG-DRIVE project was not as critical as it might be expected, therefore we sticked to the standard, water-cooled solution in terms of a commercially-available inverter. The inverter is composed of a power module which controls the machine-stator phases to produce mechanical or electrical power. The power module use depends on the voltage, either with the power Mosfet or IGBT that are driven by the control card.
The control card contains all the machine piloting algorithms in a microcontroller, as well as the rotor-position-sensor algorithms and the vehicle strategy, the battery state of charge, the machine temperature calculation, etc.
Task 7.5: Optimisation of EM power density using new permanent magnets (M24 - M30)
We selected the optimal sets of the magnets produced during the course of the project. As expected the best results were achieved by applying the HDDR+SPS route from the WP5. We were able to achieve the power densities of up to 2.56 kW/kG.
WP 8: Testing of the Magnets and Electric Motors in the Real-World Conditions (M24 - M35)
WP Leader: Jean-Marc Dubus (VALEO)
Summary of Results
This work-package was supposed to include the final research activities of the MagDrive project, since it was about testing and applying the magnets in electric motors, constructed in Valeo. It turned out to be completely successful since the delivered magnets passed all standard-prescribed tests, whereas the respective motor, performed according to specifications.
There were no deviations from the objectives and task given in Annex I.
Details for each Task
Task 8.1: Preparation for the tests (M24 – M26)
There were no special preparations needed for the tests of the magnets since all the required equipment was readily available at Valeo and Magneti. The magnetic properties were determined by applying the same techniques as described in the WP6, mainly by using the permeameter, designed for the so-called closed-loop measurements. More dedicated were the devices for the measurements of the bending force and displacement, torque, output efficiency, and thermal properties.
Task 8.2: Physical and magnetic properties of the permanent magnets (M26 – M28)
On the basis of the magnetic measurements on the powders, produced along the three routes, which were covered by the work-packages WP3, WP4 and WP5, we decided that the best magnets should result from the HDDR+SPS material. Hence we prepared bigger size magnets and cut them into smaller pieces that were appropriate for measurements and had their magnetic properties measured with permagraph at Magneti Ljubljana d.d.. Final magnets were additionally cut and ground to obtain a shape defined by Valeo, so that the diameters were about 40 mm. We ended up with 16 different magnets; four form each of the four different compacts related to the variation of the processing parameters. he results revealed a pronounced homogeneity of the compacts reflected in a reproducibility of the results. Whereas the magnets from all four compacts exhibited nearly the same remanence values of 0.8 T, the compact 1 was distinguished by the highest coercivity of 1100 kA/m, following by the compacts 3 and 4 with slightly lower values. The magnets from the compact 2 had about 10% lower coercivity, thus they were least applicable.
Task 8.3: Corrosion properties of the permanent magnets (M26 – M28)
The standard procedure in Valeo to check the corrosion resistance of electric components is the salt-spray test. It makes us possible to check the ring/brushes-compartment tightness and the vulnerability of the machine face to the electro-corrosion.
Test conditions:
Machine in rotation mounted in a specific angular position in a specific bench, with representative customer regulator plugs & an electrical circuit with a battery of 70 Ah, wiring of 35 mm2 2 m + 2 m,
Cycle: 75 min (60 min running on alternator mode (ON), 15 min stopped (OFF)) Phase « On »: Ambient temperature (Ta) Salt mass concentration Pulverization pressure Salted water quantity Rotation frequency (n) Alternator output (It) Alternator or system voltage (UB+)
Phase « Off »: Electric polarization Salt fog pulverization
: +35°C, : 5%, : 0.1 Mpa, : 1.5 ml/h, : 6 000 min-1,
: I /3 at 4 000 min-1/ +23°C, : Ureg,
: without, : without.
Cycles Number: 77 (equivalent of 96 h), For liquid cooled alternator, no liquid flow & electrical output defined in order to have Tj diode < Tj maxi.
The MagDrive magnets turned out to be suitable for the application, there was no magnet degradation, the test was passed successfully, and there were no performance losses.
Task 8.4: Tests on the new electric motor (M28 – M34)
The final tests were devoted to the evaluation of the performance of the electric-motor containing the MagDrive magnets. The measurements were carried out at different working temperature and rotation speeds. The most relevant output values were the torque and power, which were compared to the predictions of the simulations. The rotor equipped with the MagDrive magnets fulfilled the specifications. There was no demagnetization, the correlation between the simulation and measured results was good, hence all the objectives were achieved. The conclusion was that the MagDrive magnets, with reduced amount of heavy rare-earth elements, can be used in electric motors.
Potential Impact:
Dissemination and exploitation of project results are very important aspects of all European projects in order to present scientific and technological achievements to peers, financiers, industry as well as to the general public. The first step towards a better recognition of the intermediate and final results of any scientific activity is the dissemination on various levels. During the course of the MagDrive project the activities could be followed on the web page: http://mag-drive-fp7.eu. Besides a brief description of the project, its organization and the main objectives, it supplied the information about the current activities like the partners meetings and the workshops organized by the members of the consort, which were covered thematics related to the scope of the project. The workshops were open to the public and often attended by researchers or Ph.D. Students, particularly from the host institutions or companies. Those events served their purpose by spreading the information about the MagDrive project and about the EC funding scheme in general. In order to present the project in a more popular way, maybe targeting a younger generation of the future scientists, economists as well as end users, we established also a related Facebook page, where we published not just about the MagDrive but also general news from the field of magnetic materials, electric vehicles and similar topics. Of course, the scientific outcome of the project was presented in a more common way too. So far, we have published two peer-reviewed papers, given four invited talks, given six talks on scientific conferences, and presented four posters on scientific conferences. This sort of dissemination is still an ongoing activity and the list of publications is not closed yet since some of the papers are still under review, and some presentations are still to be given. The MagDrive activities were informally advertised during the the so-called doors-open days at the involved institutions. For example, at the Jozef Stefan Institute, where the project was coordinated, the doors-open days take place in March every year, attracting several thousands visitors of all ages from nursery children to seniors. It is of vital importance to inform the public about the research in the frame of the projects like MagDrive, which have a direct impact on environment (electric vehicles) and economy (lowering of the production costs- heavy-rare-earth-free magnets).
While the benefits of dissemination might be regarded as an investment for the future, the exploitation is expected to have an almost immediate impact. On the basis of the fruitful collaboration during the course of the MagDrive project, Valeo and Kochanek Entwicklungsgesellschaft initiated a new industrial project that is planed to be profitable for both partners. The improvements of the spark-plasma-sintering technique, which were developed and tested for the purposes of the MagDrive project, the patent applications “Flash Spark Plasma Sintering of fully dense, nanostructured permanent magnetic materials” was prepared by Queen Mary, University of London and University of Birmingham. This method, in a combination with the hydrogen-decrepitation-deabsorption-recombination (HDDR) process, was found to be most upscalable among the considered routes for a massive production of permanent magnets. Hence we prepared a respective detail exploitation plan, where we first compare the lab-scale and industrial-level production.
Lab-scale production
The lab-scale production of large FSPS NdFeB compacts is carried out along the following route:
1. Manual preparation of graphite die with raw material, taking 5 minutes
2. Loading die into SPS, pumping to vacuum and performing SPS on the prepared die to make pre- compact. Cooling & removal of die, taking 12 minutes
3. Manual removal of pre-compact from die and removal of graphite foil from compact, taking 5 minutes
4. Manual loading of new tooling and pre-compact into the SPS, pumping SPS to vacuum and heating to pre-heat temperature, taking 5 minutes
5. Pre-heating of compact, Flash-Sintering and cooling, taking 6.2 minutes
6. Manual removal of compact from SPS and removal of graphite foil from compact, taking 5 minutes
The overall required time is 38.2 minutes. At this rate it should be possible to produce 12 compacts during an 8 hour shift. Each compact can then be cut to produce 4 magnets. During the Mag-Drive project, the cutting of the compacts to specified dimensions was performed by external contractors. The rate of production of finished magnets from the compacts was roughly one magnet per hour. Hence, at lab scale using existing tooling, magnet production by FSPS takes approximately 1hr 40min per magnet.
Based on this, it is clear to see that the rate of production is limited by numerous factors, including: All manual preparation of dies and unloading/cleaning of sintered compacts. Pumping the furnace to vacuum. Heating the furnace to the sintering or pre-heat temperature.
Waiting for the die/compact to cool before the next die/compact can be loaded. Cutting and finishing of magnets from the compacts.
In addition, the cutting operation leads to the loss of around 50 % of the compact as scrap; since rectangular magnets are being cut from a circular disk of material and since the edges of the compacts need to be avoided due to cracking and increased porosity in these areas. The cracking and porosity at the edges of the compacts is due to the increased loss of heat from these regions during FSPS. Graphite felt is placed around the edges of the compact to reduce this effect but due to the lack of control over the temperature of the felt, this does not eliminate the problem all together.
Scaling up to industrial level
Based on the above assessment of lab scale production: a continuous, automated process to produce net-shape forged magnets would be the most cost-effective and resource-efficient method to work towards in the scale-up to industrial level. Lab scale FSPS experiments with Nd-Fe-B have shown that very large and rapid degrees of deformation can be achieved using only small applied pressures. This highlights the potential for efficient net-shape forging by FSPS; provided that the die used is electrically insulated from the FSPS tooling and is heated. This subsequently avoids the uneven heating and surface cracking problems associated with the lab scale process. The proposed processing route consists of four steps:
1. Filling of pre-press die by automatic hopper system, taking two minutes
2. Cold pressing, taking eight minutes
3. IR ‘biscuit’ firing & preheating, taking 60 minutes
4. Net-shape FSPS, taking 10 minutes
The estimated total time is 80 minutes without incorporating cooling into the “per-magnet” processing time, since multiple magnets can be left to cool at the same time; hence this is not a rate-limiting step.
The continuous aspect of the process ensures a high throughput of material. Since a vacuum is maintained at all times and there is no waiting for the chamber to reach the pre-heat temperature, the time and energy for processing is cut considerably. Automation minimises the time and expense from manual labour operations. Using cold compaction and biscuit firing instead of a pre-SPS step to produce the green compact reduces the pre-compact fabrication time by around 21.5 min; particularly due to the fact that die preparation and the removal of graphite foil from the sample are no longer necessary. Finally, the net shape forging step removes the need for post-fabrication machining, which significantly reduces the magnet fabrication time and eliminates raw material waste all together. As shown above, the total time for the fabrication of each magnet using this process, is 80 seconds. This would equate to the production of 360 magnets per day, based on one 8 hr shift.
Of note is that net-shape forming also provides added value to the finished magnets, particularly if more complex shapes could be formed. Currently, complex-shape magnets are made from ferrite materials which have been machined into the final shape. The machining process results in a lot of material loss and is therefore only economically viable when applied to cheap materials. If net- shape materials could be produced without any material wastage, then it would become economically viable to produce complex-shape Nd-Fe-B magnets. This would be advantageous in e.g. electric motors; where the volume of magnetic material required to produce the same power as the equivalent ferrite-based machine would be roughly halved. Smaller, lighter and more powerful machines would then be possible.
Continuous, multi-chamber, ‘tunnel type’ SPS machines are already in existence, manufactured, for example, by NJS Co. Ltd.
The figure of 360 magnets per 8 hr day is based upon a single line of production; with only one magnet being Flash Sintered at any one time. There is no reason why the SPS machine and tooling could not be developed so that several magnets could be simultaneously Flash Sintered (using e.g. multi-cavity cold pressing and FSPS tooling systems), multiplying the machine output. 6-10 cavity dies have already been employed in regular SPS. Additionally, back-to-back shift work could be introduced to double the daily output. With this in mind, the machine and tooling could be developed to produce a throughput of anything between 360 and 7200 magnets per day.
As demonstrated, the upscaling of the processing procedure is feasible. In the following we focus on the cost analysis, which is divided into the cost of the industrial FSPS, and into the cost of the raw material.
Cost analysis of industrial scale FSPS
In consultation with Kennametal UK Ltd., who own and run a single-chamber industrial scale SPS, costs have been drawn up for the purchase and running of an industrial scale continuous SPS.
The cost of a new, customized, multi-chamber SPS as discussed above is roughly EUR 1.74M. For the FSPS process suggested, an additional chamber for the hopper loading and cold pressing of the Nd-Fe-B powder would have to be developed and added to the machine; at an estimated maximum cost of EUR 580k. The investment required for the research and development of the custom machine would be roughly EUR 232k. With attendant costs of building, services and other incidental one-off expenses, the total one-off cost for the customized SPS machine and installation is expected to be around EUR 2.9M:
Machine research and development investment: EUR 232k
New multi-chamber SPS furnace and associated equipment, in accordance with FSPS requirements and desired throughput: EUR 2.74M
Building / Services to accommodate new SPS facility: EUR 580k
Purchase of handling aids to help in periodic removal of large numbers of magnets from the machine and to refill the hopper with powder: EUR 232k
Ongoing running costs associated with the SPS include:
Part of the time of a trained technician to periodically load the hopper with powder and remove cooled magnets from the machine. Considering two technicians working back-to-back for overnight production, this could equate to a pro rata cost of EUR 35k per annum.
Replacement of custom made FSPS graphite tooling and purchase of different sets for different desired magnet shapes. For the proposed net shape FSPS process this only relates to the punches being used. Given the relatively low pressures required for the small magnets the punches are expected to last for up to 200 runs. It is therefore recommended that the punches be changed twice a day. Considering the highest output case of the multi-processing of 10 magnets at once, this equates to 100 punch sets a week and 5200 punch sets in a year. The cost for one set of custom designed punches is EUR 58, hence this is an ongoing expense of EUR 302k per year. Potential savings could be made by producing the punch sets in-house.
Ongoing SPS maintenance and repairs, EUR 23-35k per year.
Energy bills for running of the SPS machine. The FSPS process consumes around 0.02 kWh per magnet, which at a maximum throughput of 7200 magnets per day, would equate to 144 kWh per day or 37,440 kWh per annum. Factoring in other associated energy costs (e.g. the intermittent running of a vacuum pump and chiller), the annual energy consumption is expected to be roughly 70,000 kWh/annum; costing around EUR 8k:
Technician time: estimated EUR 34.7k per year
Graphite tooling: estimated EUR 300.5k per year
SPS maintenance and repairs: estimated EUR 23-34.7k per year
Energy bills: estimated EUR 8.1k per year
which sum up to the estimated total cost of maximum EUR 377.9k per year.
Raw materials cost
The current cost to make and ship a standard 12g Nd-Fe-B magnet containing 10 wt.% Dy is around EUR 1.2. This price is based on analysis of the current Chinese local market and cost when supplied to Europe, as determined in consultation with Valeo. Of this figure, the raw materials cost accounts for EUR 1.056.
For a Dy-free composition, as per the current MQU-F material being processed by FSPS, the raw materials cost would be EUR 0.4 per magnet. For a maximum output of 7200 magnets per day, this equates to an annual raw materials cost of around EUR 760,000.
The above assumptions lead us to the final estimate.
Production cost per magnet and & break-even point
Taking into account the ongoing and raw materials costs, and neglecting the initial one-off costs, the cost of production per magnet equals EUR 0.608. If the magnets were retailed at a competitive market price of EUR 1.20 each plus VAT, this would equate to a profit of EUR 0.592 per magnet and EUR 1,108,224 per annum.
The break-even point, at which the initial one-off investment costs equal the profit, would therefore be expected to occur at the end of the third quarter of the third year after starting the above-described industrial production.
In summary, We demonstrated that the transfer of the spark-plasma sintering (SPS) method for the production of permanent magnets without heavy-rare earth content, which was developed during the course of the MagDrive project, could be commercially successful, which is the single most exploitation achievement of the project.
An eventual single company, based on the above-described process, would offer several hundreds new working positions within EU with a possibility for further growth, depending on the market demands. These will become big, if the vision of electric vehicles dominating the roads, as promoted by the MagDrive project, comes true.
List of Websites:
http://mag-drive-fp7.eu
