Rare Earths and Magnetic Refrigeration

Acknowledgement: This work was supported by the U.S. Department of Energy, Office of Basic Energy Sciences, Materials Sciences and Engineering Division; and Astronautics Corporation of America, Milwaukee, Wisconsin.

*Corresponding author (E-mail: cagey@ameslab.gov)

Abstract: Magnetic refrigeration is a revolutionary, efficient, environmentally friendly cooling technology which is on the threshold of commercialization. The magnetic rare earth materials are utilized as the magnetic refrigerants in most cooling devices, and for many of the cooling applications the Nd2Fe14B permanent magnets are employed as the source of the magnetic field. The status of near room temperature magnetic cooling is reviewed.

Key words: magnetic refrigeration; magnetocaloric effect; gadolinium; Gd5(Si1-xGex)4; La(Fe13-xSix)Hy; Nd2Fe14B permanent magnets; active magnetic regenerator cycle; rare earths.

Modern cooling is almost entirely based on a compression/expansion refrigeration cycle. It is a high-energy demand industry with annual energy consumption measured in billions kWh. Over the years, all parts of a conventional refrigerator, i.e. compressors, heat exchangers, refrigerants, and packaging have been considerably improved due to an extensive research and development effort, and in part by government edicts. This was made possible by a continuous dollar influx from both federal and industrial sources. Both achieved and anticipated improvements of this traditional technology, however, are incremental since modern refrigeration is already near its fundamental limit of energy efficiency, which is well below the maximum theoretical (Carnot) efficiency. Furthermore, the liquid chemicals used as refrigerants, eventually escape into the environment promoting ozone layer depletion and global warming and, therefore, conventional refrigeration ultimately promotes deleterious trends in the global climate. Other refrigerants, such as ammonia, are hazardous chemicals.

In 1997 a new, revolutionary, competitive, energy efficient, and environmentally friendly cooling technology emerged – magnetic refrigeration (MR). Two major events ushered in this new cooling technology – first, the unveiling of a proof-of-principle working magnetic refrigerator on February 20, 1997 at Madison, Wisconsin[1], and second, the discovery of the giant magnetocaloric effect in Gd5Si2Ge2 and the related Gd5(Si1-xGex)4 alloys announced on June 9, 1997[2,3].

The February breakthrough was the successful testing of the Ames Laboratory, Iowa State University/Astronautics Corporation of America (AL,ISU/ACA) reciprocating proof-of-principle magnetic refrigerator. This machine operated in magnetic fields up to 50 kOe (5 T) provided by a superconducting magnet. It achieved a cooling power of 600 watts with a coefficient of performance (COP) approaching 15, an efficiency reaching 60% of Carnot (a ~50% improvement over a typical vapor compression system), a temperature span of 38 K in magnetic fields of 5 T, and operated for over 1500 hours without any mechanical or electrical problems.
The June breakthrough showed that there was at least one family of alloys, the Gd5(Si1-xGex)4 compounds, that might be much better refrigerants than the prototype Gd metal magnetic refrigerant because of the much larger magnetocaloric effect. This discovery not only brings magnetic.

refrigeration one step closer to commercialization, but it has spawned explosive growth of related research, the results of which indicate that these unique materials display some extraordinary magnetic properties potentially useful in other energy-related applications. These unique properties include the colossal magnetostrictive and giant magnetoresistive behaviors[4], which can be used in energy conversion devices and data storage applications. Furthermore, the discovery of the giant magnetocaloric effect spurred a broad international interest in the magnetocaloric effect and lead to the discovery of four new families, members of which exhibit the giant magnetocaloric effect. These are: Mn(As1-xSbx), MnFe(P1-xAsx), La(Fe13-xSix)Hz and Ni~55Mn~20Ga25.

Between 1998 and 2006, following the Ames Laboratory and Astronautics Corporation of America footsteps, 19 more magnetic refrigerators have been built and tested by scientists and engineers in Canada (1), China (7), Europe (4), Japan (5) and the USA (3), signaling the dawn of a new era of environmentally friendly, energy efficient and affordable magnetic cooling, refrigeration and air conditioning. The most advanced magnetic cooling machine is the laboratory prototype, permanent magnet, rotating refrigerator built by Astronautics Corporation of America in Madison, Wisconsin. It was publicly displayed at the Global Eight (G8) Energy Ministers Conference in Detroit, Michigan on May 1, 2002 and again at the Anniversary Celebration of the President’s National Energy Policy at the US Department of Energy Headquarters in Washington, DC on May 17, 2002. More information about nine of these refrigerators and the references to the original papers will be found in a 2005 review by Gschneidner, et al.[5].
The interest in magnetic cooling is continuing to grow rapidly. This can be attested to by the fact that in September 2005, the first international conference on magnetic refrigeration near room temperature was held in Montreux, Switzerland.[6] The next conference will be held in April 2007.

1 The Magnetocaloric Effect
The magnetocaloric effect (MCE) is the response of a magnetic solid to the application (or removal) of a magnetic field, which is evident by a change in the temperature of the solid. For a ferromagnetic material near its magnetic ordering temperature (the Curie temperature [TC]), when a magnetic field is applied, the unpaired 4f or 3d spins are aligned with the magnetic field, which decreases the entropy in the isothermal process or causes the sample to warm up in the adiabatic process. When the magnetic field is turned off the spins randomize increasing the entropy, or the material cools. A few materials, primarily antiferromagnetic compounds, may exhibit the opposite behavior; they cool when a magnetic field is applied, and warm up when the field is removed.

The temperature change is called the adiabatic temperature change, ΔTad. The extensive parameter representing the MCE is the isothermal magnetic entropy change, ΔSm. For Gd metal ΔTad ≅ 5.7 K and ΔSm ≅ 5.5 J/kgK (~43 mJ/cm3K) for a magnetic field change of 20 kOe (2T), see Fig. 1. Most magnetic solids undergo a second order magnetic transition when the solid orders magnetically, resulting in what has been called the conventional magnetocaloric effect.

A few magnetic materials, however, exhibit a significantly larger MCE, which is known as the giant magnetocaloric effect (GMCE). It occurs because the magnetic solid undergoes a coupled first order magnetic-structural transition. The ΔSm for a GMCE material may be twice or more as large as the ordinary MCE of a substance which undergoes a second order transition. The extra entropy is due to the entropy difference between the two structures involved in the transition. However, since the GMCE materials undergo a first order transition the very nature of this transition means that it often exhibits hysteresis and time dependence, and these may limit the usefulness of the GMCE materials in magnetic refrigeration. More details about the MCE can be found in Refs. [7-11].

2 Magnetic Refrigeration
The basic principle of magnetic cooling is analogous to gas compression cooling. In order to achieve continuous refrigeration, one must reject the heat generated in the magnetic material to the ambient when the magnetic field in increased. This is equivalent to the compression stage of a gas cycle refrigerator or air conditioner. During the demagnetization step, a thermal link is used to cool the load which is equivalent to the gas expansion stage of a compressor. By continued repetition of these two steps refrigeration is accomplished.

2.1 Regenerators
One can increase the performance of the magnetic cooling device considerably by simultaneously using the magnetic refrigerant as a regenerator. A regenerator is a thermal device that absorbs heat from a fluid heat transfer medium (a gas or liquid) as it passes through the regenerator thereby cooling the fluid. The cooled fluid absorbs heat from the item to be cooled and on the reverse portion of the cycle the fluid passes back over the regenerator material extracting heat to a hot heat exchanger which rejects the heat to the ambient surroundings. In general, the higher the volumetric heat capacity of the regenerator material the more efficient the cooling device.
The regenerator is constructed of either a bed of tightly packed spheres (or irregularly shaped powders), or a set of closely spaced parallel thin plates (foils), or a wire mesh. The size of the spheres vary from 100 to 300 μm to provide effective heat transfer between the regenerator materials and the heat transfer fluid, and sufficient porosity to minimize the pressure drop across the bed of spheres. The plate (foil) thickness and wire mesh diameter are comparable to sphere diameter for the same reasons.

2.2 Active magnetic regenerator cycle[5,12-14]
In the active magnetic regenerator (AMR) cycle a porous bed of a magnetic refrigerant material acts as both the refrigerant (coolant) and the regenerator for the heat transfer fluid. Assume that the bed is at steady state with the hot heat exchanger at ~24°C and the cold heat exchanger at ~5°C (dashed line in Fig. 2(a)). The first step in the AMR cycle is to apply a magnetic field to the refrigerant, each particle in the bed warms because of the MCE to form the final magnetized bed temperature profile (solid line in Fig. 2(a)). The amount each particle warms is equal to ΔTad reduced by the effect of the heat capacity of the heat transfer fluid in the pores between the particles. The second step in the cycle is to push the ~5°C fluid through the bed from the cold end to the hot end. The bed is cooled by the fluid lowering the temperature profile across the bed (the dashed line to the solid line in Fig. 2(b)) and the fluid in turn is warmed by the bed, emerging at a temperature close to the temperature of the bed at the warm end. This temperature is higher than ~24°C, so heat is removed from the fluid at the hot heat sink as the fluid flows through the hot heat exchanger. After the fluid flow is stopped, the magnetic field is removed (step three), cooling the bed by the MCE (the dashed line to the solid line in Fig. 2(c)). The refrigeration cycle is completed by forcing the 24°C fluid to flow from the hot to the cold end of the bed (the dashed line to the solid line in Fig. 2(d)). The fluid is cooled by the bed, emerging at a temperature below ~5°C and removes heat from the cold sink as the fluid passes through the cold heat exchanger.

The AMR cycle outlined above has several positive features useful for practical application in a magnetic cooling device. One, the temperature span of a single stage can greatly exceed that of the MCE of the magnetic refrigerant because the MCE of each individual particle changes the entire temperature profile across the bed. Two, because the bed is a regenerator, heat need not be transferred between two separate solid assemblies, but rather between the solid particles in a single bed via the action of a fluid. Three, the individual particles in the bed do not encounter the entire temperature span of the stage, and hence the bed may be made into layers, each containing a magnetic material with properties optimized for a particular temperature range.

2.3 Regenerator materials
Gadolinium metal is considered the prototype magnetic refrigerant material for near room temperature magnetic refrigerators. It is a good refrigerant, but to make magnetic refrigeration even more efficient it is important to have an array of materials with better MCE properties than Gd. Of course, there are other ways to increase the efficiency, such as new and better thermodynamic cycles, improved engineering designs, increase the number of AMR cycles per unit time, better heat transfer fluids. Since the discovery of the GMCE in the Gd5(Si1-xGex)4 alloys[2,3] hundreds of magnetic materials with magnetic ordering temperatures ranging from 1 to 400 K have been reported in the literature.[4,15-18] The other families of compounds/alloys which have at least one composition with a GMCE near room temperature include: the rare earth manganites, (R1-xMx)MnO3 where R = a rare earth metal, and M = Ca, Sr or Ba[5,15]; Mn(As1-xSbx) alloys[19]; MnFe(P1-xAs) alloys[20]; the Ni~2Mn~1Ge~1 Heusler alloys[21]; La(Fe13-xSix) materials[22,23]; and La(Fe13-xSix)Hy alloys.[23] In most of the cases the basis for claiming a GMCE is the large ΔSm value which is calculated from magnetization measurements. However ΔTad is also an important parameter for the successful operation of a magnetic cooling machine. Most of these new materials have small ΔTad values for a given magnetic field change when compared to Gd metal, the only exception is the Gd5(Si1-xGex)4 alloys for a 0 to 50 kOe field change, ΔTad is ~40% larger than that of Gd. However, for a 0 to 20 kOe field change La(Fe13-xSix)Hy and MnFe(P1-xAsx) have ΔTad values which are comparable to that of Gd, as well as the Gd5(Si1-xGex)4 alloys.

There are, however, a few potential problems for first order magnetic materials which may limit their usefulness as magnetic regenerator materials – hysteresis[24], the temperature and magnetic field ranges over which the transformation occurs, and the time it takes to reach the full ΔTad value.[5] If the hysteresis is quite large the original magnetic phase may not be fully recovered during a heating-cooling cycle, and thus on the next cycle one does not get the full MCE because part of the magnetic refrigerant was already in the high magnetic field state. One can get around this problem by correctly designing the AMR cycle and carefully layering the magnetocaloric bed with a range of materials with various TCs such that each particle along the bed has its optimum MCE at steady state.

If the magnetic-structural transformation takes place over a wide temperature range (2 to 10 K) and a wide magnetic field range (2 to 10 kOe), only part of the original magnetic phase may have transformed. Grain size and sample purity probably play important roles on the width of the transformation, but this problem has not been studied in any detail.

Recently it was pointed out [5] that there is a time dependence in the measurements of the ΔTad– the quicker the measurement the smaller the value of ΔTad. This can present a real problem in magnetic refrigeration because the cycle frequencies range from 0.2 to 4 Hz. This is only true for first order magnetic-structural transitions, since there is a movement of atoms during the transition, volume change and related strain, while there is no time dependence for second order transformations because they only involve alignment (or disalignment) of the magnetic 4f or 3d electrons.

2.4 Evaluation of regenerator materials
There are a number of other important criteria for materials selection of a magnetic refrigerant for an operational magnetic cooling device in addition to have good to excellent ΔSm and ΔTad values. These include the raw material cost, the preparation of ton per day quantities, the vapor pressure of the components, fabrication costs to form the material into a useful form for the regenerator, the refrigeration capacity (i.e. how much heat can be transferred per cycle), hysteresis, time dependence of ΔTad, environmental concerns, and corrosion. These have been discussed in some detail and the various highly touted magnetic refrigerant materials have been compared to Gd metal.[5] As a whole, there is no material to date which clearly is better than Gd (and Gd doped with other rare earths). But more research and development needs to be carried out to see if some of the potentially adverse problems of other families of magnetic materials can be solved. Until that day Gd is still the material of choice as the magnetic refrigerant.

3 Magnetic Field Source
The strength of the magnetic field is important in the utilization of the MCE in magnetic cooling, since both the ΔSm and ΔTad are approximately proportional to the magnitude of the magnetic field change.[5] At low field, <20 kOe, the two parameters vary nearly linearly with the magnetic field, but at fields >20 kOe the MCE change per unit magnetic field change becomes somewhat smaller as the field increases, i.e. the slope deviates from the linearity established at lower fields.

Because of this dependence the efficiency of magnetic cooling increases with increasing field.[5] Thus one wants as large a magnetic field as possible in the magnetic cooling device. However, there are some practical considerations which need to be taken into account. For example, a superconducting magnet can easily be designed to provide a magnetic field of 50 to 70 kOe (5 to 7 T) at a reasonably modest cost, but such a magnetic field source is impractical for household applications and most transportation methods.
The only practical magnetic field source for household and transportation applications is the high energy Nd2Fe14B permanent magnet which can deliver a magnetic field of 12 to 15 kOe for a reasonable gap size between the pole pieces to allow one to move the magnetic refrigerant material in and out of the magnetic field. Applications which would utilize the rare earth permanent magnets are: household refrigerator/freezers; household air conditioning; automotive, aircraft, small seafaring vessels for climate control (both heating and cooling, and humidity); portable coolers; active cooling of electronics; portable refrigerators (medical).

The use of a superconducting magnet source is the only practical way for large scale applications, such as supermarket chillers; frozen food plants; building climate control (heating, cooling and humidity) such as office buildings, apartments, convention centers, meeting halls; and climate control for large seafaring vessels. As far as we are aware no one has designed, built and tested a large scale cooling machine or even a prototype apparatus.

4 Impact on Rare Earth Markets
Magnetic refrigeration will have a tremendous impact on the rare earth markets as this technology grows and matures. We believe that the first commercial units will be on the market within a few years. How rapidly this market will grow is difficult to predict. In part it will depend upon the speed at which some of the impediments to commercialization, see below, are overcome. The amount of magnetic refrigerants required will range from 0.01 to 100 kg per cooling device. The smaller quantity would be for small refrigerators and coolers noted above, while the larger kilogram quantities are required in the magnetic cooling machines for large scale applications.

For refrigerators which will use permanent magnets as the power source we believe 0.5 to 100 kg of the Nd2Fe14B alloy are required per refrigerator or cooling device.

Thus magnetic refrigeration, even before it matures, will be the biggest rare earth market. This is especially true for the small scale units since both the magnetic refrigerant material and the permanent magnet will contain significant amounts of rare earth metals per cooling unit.

It is conceivable that in ten years from now the high magnetic field source will be a superconducting magnet which uses one of the high temperature ceramic superconductors, such as YBa2Cu3O7. If this should occur it will be another big plus to the rare earth market. In addition the use of the ceramic oxide superconductor is expected to reduce the operating costs compared to the conventional liquid helium superconducting system in use today.

5 Obstacles to Commercialization
As noted above at least 20 laboratory scale prototype magnetic refrigerators have been built and tested, but there are still a number of obstacles which need to be overcome for successful large scale commercialization. The main obstacles are as follows, but they are not listed in any priority.

Magnetic refrigerants, which have a second order magnetic transition, with better MCE properties than Gd would give a big boost towards moving this technology to large scale production of magnetic cooling devices. This is especially true if the costs associated with forming the final regenerator material (spheres, wires, foils) are not much more expensive than it costs to make Gd foils and/or spheres and/or wire mesh.

The production of large quantities of the magnetic refrigerant, tons per day at a reasonable cost is critical. Most of the materials studied to date have only been prepared in small quantities, generally less than 25 g. To date only one material, the Gd5(Si1-xGex)4 family of alloys, has been successfully produced at a kilogram scale.[5]

Permanent magnets which have higher magnetic strength and lower costs will benefit the commercialization. As much as one-third of the cost of a magnetic refrigerator is the magnet source. Improved magnetic field strength will lead to high efficiencies, and this will allow the engineer to reduce the size of the permanent magnet array for the same amount of cooling.

Engineering considerations should be examined closely to improve the efficiency. These improvements could involve: the heat flow from the cold object to the hot heat exchanger; the heat exchanger themselves; increasing the cycle frequency; and the heat exchange fluid, etc.

The AMR thermodynamic cycle is considered to be the best cooling cycle today for magnetic cooling/heating, but it is possible by making some modifications it could be improved. New thermodynamic cycles, especially to take into account the MCE behaviors of first order magnetic transitions, would be a big step forward to utilize their unique properties in a practical magnetic refrigerator.

The first order magnetic transformation materials have a large MCE (i.e. GMCE) due to the fact that in addition to normal magnetic entropy associated with magnetic ordering, there is a second contribution, the entropy associated with the structural change in materials which exhibit a magnetic-structural transformation. The magnitudes of the two entropies are comparable. There are several problems associated with such material, one of which is the hysteresis inherently associated with first order transitions. It may be possible to reduce the hysteresis, but never eliminate it, by increasing the purity, and/or grain size, and perhaps by alloying an appropriate impurity element. Also as noted above by carefully packing regenerator beds with the appropriate alloy compositions from the cold to hot ends could circumvent any problems due to the hysteresis. Finally there appears to be a time dependence associated with these first order transformations, on the order of minutes to realize the full MCE, especially ΔTad. This problem may be more critical than hysteresis if first order magnetic transition materials are to be used..

6 Conclusions
Commercial magnetic refrigeration/cooling units will be a reality in the near future. But whether or not it will be just a niche market, or a full-blown growth market 5 to 10 years from now is difficult to predict. Either way this technology will be an important market for the rare earth industry.

References
[1] Zimm C, Jastrab A, Sternberg A, et al. Description and performance of a near-room temperature magnetic refrigerator [J]. Cryog. Eng., 1998, 43: 1759.
[2] Pecharsky V K, Gschneidner Jr K A. Giant magnetocaloric effect in Gd5(Si2Ge2) [J]. Phys. Rev. Lett., 1997, 78: 4494.
[3] Pecharsky V K, Gschneidner Jr K A. Tunable magnetic regenerator alloys with a giant magnetocaloric effect for magnetic refrigeration from ~20 to ~290 K [J]. Appl. Phys. Lett., 1997, 70: 3299.
[4] Pecharsky V K, Gschneidner Jr K A. Gd5(SixGe1-x)4: an extremum material [J]. Adv. Mater., 2001, 13: 683.
[5] Gschneidner Jr K A, Pecharsky V K, A O Tsokol. Recent developments in magnetocaloric materials [J]. Rept. Prog. Phys., 2005, 68: 1479.
[6] Egolf P W. ed. 1st International Conference on Magnetic Refrigeration at Room Temperature [A], Paris: Institut International du Froid, 2005.
[7] Pecharsky V K, Gschneidner Jr K A. Magnetocaloric effect and magnetic refrigeration [J] . J. Magn. Magn. Mater., 1999, 200: 44.
[8] Tishin A M, Gschneidner, Jr K A, Pecharsky V K. Magnetocaloric effect and heat capacity in the phase-transition region [J]. Phys. Rev. B, 1999, 59: 503.
[9] Pecharsky V K, Gschneidner Jr K A. Magnetocaloric effect from indirect measurements: magnetization and heat capacity [J]. J. Appl. Phys., 1999, 86: 565.
[10] Pecharsky V K, Gschneidner Jr K A. Heat capacity near first order phase transitions and the magnetocaloric effect: an analysis of the errors, and a case study of Gd5(Si2Ge2) and Dy [J]. J. Appl. Phys., 1999, 86: 6315.
[11] Pecharsky V K, Gschneidner Jr K A, Pecharsky A O, Tishin A M. Thermodynamics of the magnetocaloric effect [J]. Phys. Rev. B, 2001, 64: 144406.
[12] Barclay J A. The theory of an active magnetic regenerative refrigerator. Proceedings of the Second Biennial Conference on Refrigeration for Cryocooler Sensors and Electronic Systems [C], NASA-CP-2287, Greenbelt, MD; Goddard Space Flight Center, 1983, also available as a Los Alamos National Laboratory report LA-UR-82-1792.
[13] Barclay J A, Steyert W A. Active magnetic regenerator [P]. U.S. Patent 4,332,135 (June 1, 1982).
[14]Zimm C B, DeGregoria A J. Magnetic refrigeration: application and enabler for HTSC magnets. In Kwok H S, Show D T, Naughton H J, eds. Proceedings of the 6th International Conference on Superconducting Applications, AIP Conf. Proc. 273 [A], New York: American Institute of Physics 1993, 471.
[15] Gschneidner Jr K A, Pecharsky V K. Magnetocaloric materials [J]. Ann. Rev. Mater. Sci., 2000, 30: 387.
[16] Gschneidner Jr. K A, Pecharsky V K. Magnetic refrigeration. In Westbrook J H, Fleischer R L, eds. Intermetallic Compounds: Vol. 3, Principles and Practice [M]. New York: John Wiley & Sons, Ltd., 2002, 519.
[17] Tishin A M, Spichkin Y I. The Magnetocaloric Effect and its Applications [M]. Bristol: Institute of Physics Publishing, 2003.
[18] Brück E. Developments in magnetocaloric refrigeration [J]. J. Phys. D: Appl. Phys., 2005, 38: R381.
[19] Wada H and Tanabe Y. Giant magnetocaloric effect of MnAs1-xSbx [J]. Appl. Phys. Lett., 2001, 79: 3302.
[20] Tegus O, Brück E, Buschow K H J, de Boer F R. Transition-metal-based magnetic refrigerants for room-temperature applications [J]. Nature, 2002, 415: 150.
[21] Albertini F, Canepa F, Cirafici S, Franceschi E A, Napoletano M, Paoluzi A, Pareti L, Solzi M. Composition dependence of magnetic and magnetothermal properties of Ni-Mn-Ga shape memory alloys [J]. J. Magn. Magn. Mater., 2004, 272-276: 2111.
[22] Hu F-X, Shen B-G, Sun J-R, Cheng Z-H, Rao G-H, Zhang X-X. Influence of negative lattice expansion and metamagnetic transition on magnetic entropy change in the compound LaFe11.4Si1.6 [J]. Appl. Phys. Lett., 2001, 78: 3675.
[23] Fujita A, Fujieda S, Hasegawa Y, Fukamichi K. Itinerant-electron metamagnetic transition and large magnetocaloric effects in La(FexSi1-x)13 compounds and their hydrides [J]. Phys. Rev. B, 2003, 67: 104416.
[24] Provenzano V, Shapiro A J, Shull R D. Reduction of Hysteresis losses in the magnetic refrigerant Gd5Ge2Si2 by the addition of iron [J]. Nature, 2004, 429: 853; and corrigenda 430: 810.