Recent developments in room temperature Active Magnetic Regenerative Refrigeration.
Active magnetic regenerative refrigeration (AMRR) systems represent an environmentally attractive alternative to vapor compression systems that do not use a fluorocarbon working fluid. The AMRR concept has previously been demonstrated using superconducting solenoid magnets that are not practical for small-scale commercial applications. However, recent AMRR prototypes that use more practical permanent magnets have proved that AMRR systems can produce cooling over a useful temperature range with a relatively low magnetic field. In addition, families of materials with large magnetocaloric effects and adjustable Curie temperatures have been developed; these materials may be used to construct layered regenerator beds that may have lower cost and provide higher performance than current materials. This paper reviews recent developments in the field of room temperature magnetic refrigeration and discusses some design issues that may affect practical systems.
Active magnetic regenerative refrigeration (AMRR) systems represent an attractive alternative to vapor compression refrigeration and air-conditioning systems. AMRR systems do not use a fluorocarbon working fluid; instead, a solid refrigerant is used. The solid refrigerant, a magnetocaloric material, communicates with the environment via a heat transfer fluid. Because the solid refrigerant has essentially zero vapor pressure, AMRR systems have no ozone depletion potential (ODP) and no direct global warming potential (GWP). The heat transfer fluid will likely be aqueous and will therefore have minimal environmental impact. In theory, a well-designed AMRR system can be competitive with or even more efficient than vapor compression systems, provided that the volume of the active magnetic regenerator is sufficiently large.
The Thermodynamics of the AMRR Cycle
The temperature and magnetic field of a magnetocaloric material are highly coupled over certain, typically limited, operating ranges; this characteristic allows them to be used within energy conversion systems. A thermodynamic substance can change its internal energy (U) as a result of either work or heat, leading to the differential energy balance:
dU = TdS + dW (1)
The first term in Equation 1 corresponds to an inflow of heat (TdS) and the second to an inflow of work (dW). In general, work can flow in many forms (e.g., mechanical, electrical, etc.). The familiar fundamental property relationships that describe most fluids result when only volumetric compression work (P-V) is considered; however, for magnetocaloric materials, the [[mu].sup.0]H) and magnetic moment (M) form the work term in Equation 1 hysteresis effects are ignored (Guggenheim 1967).
dU = TdS + [[mu].sup.0]HdM. (2)
Increasing the applied field for magnetic materials tends to align the magnetic dipoles, which requires work and reduces entropy. Using this relationship between entropy, internal energy, and magnetic field, it becomes possible to apply all of the typical thermodynamic results and identities that are ordinarily used in the context of a pure compressible substance to a magnetocaloric material. For example, Maxwell's relations (Guggenheim 1967) can be used to describe relationships between the partial derivatives of properties, and a magnetocaloric material will be characterized by an equation of state that describes the magnetization as a function of temperature and applied field. A temperature-entropy diagram for a magnetic material will include lines of constant applied magnetic field rather than isobars; however, the diagram is otherwise analogous to a more familiar diagram characterizing a compressible working fluid. For example, Figure 1 illustrates the temperature-entropy diagram for an alloy of 94% Gadolinium and 6% Erbium, G[d.sup.[0.94]]E[r.sup.[0.06]] (Zimm et al. 2003).
Closer examination of Figure 1 reveals that it is possible to change the temperature of a magnetic material in an adiabatic process by changing the applied magnetic field. Figure 2 illustrates the adiabatic temperature change of G[d.sup.0.94]E[r.sup.[0.06]], when the magnetic field is increased from 0 to 2 Tesla and from 0 to 5 Tesla. Figure 2 shows that the adiabatic temperature change (which is a direct indicator of the magnetocaloric effect) depends on the initial temperature of the material and that a large magnetocaloric effect is only exhibited for a relatively limited temperature span. In a material such as G[d.sup.[0.94]]E[r.sup.[0.06]] that exhibits a second-order phase transition above the magnetic ordering temperature, magnetic hysteresis does not exist. In general, magnetic anisotropy goes to zero when approaching the magnetic ordering temperature. In this case, adiabatic magnetization and demagnetization are isentropic processes; therefore, when the material is subsequently demagnetized, its temperature will return to its original, zero-field value.
[FIGURE 1 OMITTED]
Figure 2 reveals several details that are relevant to practical AMRR systems. First, the adiabatic temperature change is relatively small compared to the temperature span required for most practical cooling systems. This characteristic necessitates the use of a regenerative cycle in order to provide a cooling load over a useful temperature span. Second, the magnetocaloric effect is largest over a relatively narrow temperature range. In order to maximize the magnetocaloric effect and therefore the performance of the AMRR system, it is desirable to construct a regenerator bed from several materials that have Curie temperatures that are tailored to the local regenerator temperature.
[FIGURE 2 OMITTED]
Magnetic Cooling System Configurations
Early magnetic coolers were used to achieve extreme cryogenic temperatures and used an adiabatic demagnetization refrigeration (ADR) cycle. Giauque and MacDougall (1933) used an ADR system to reach temperatures below 1 K, breaking the temperature barrier that had previously been imposed by the properties of compressible fluids. The ADR system that they and other researchers used consisted of a solid piece of magnetocaloric alloy that utilized an isothermal magnetization in which the material is placed into contact with a hot reservoir followed by an adiabatic demagnetization. All of the material in an ADR cycle undergoes the same thermodynamic cycle and therefore the temperature lift is limited to the adiabatic magnetization temperature change exhibited by the material. ADR cycles also require complex heat switches with limited capacities. For these reasons, the ADR cycle is not a practical alternative for near room temperature, commercial devices.
The technical barriers associated with the ADR cycle have been overcome by the use of a regenerator within the active magnetic regenerative refrigeration (AMRR) cycle. Brown (1976) first constructed a regenerative magnetic refrigerator and showed that the use of a regenerative configuration can provide a no-load temperature span that is much greater than the adiabatic temperature change of the magnetocaloric material that is used to construct the regenerator. Green et al (1986) constructed the first successful AMRR, which achieved a 40 K temperature span. In an AMRR system, a porous bed of magnetic material is exposed to a time-varying magnetic field and a time-varying flow of heat transfer fluid. Each segment of the bed undergoes a unique refrigeration cycle and interacts with the adjacent material via the heat transfer fluid. The net result of these cascaded refrigeration cycles is a temperature lift that is much larger than can be achieved by an ADR cycle.
The AMRR cycle consists of four processes. A conceptual drawing illustrating the processes that make up the operation of a rotary AMRR, such as is described by Zimm et al. (2006), is shown in Figure 3. A regenerator consisting of six individual beds is discussed here; one of the six beds is highlighted in Figure 3 and is considered in the following discussion. The bed is magnetized by rotating it into the field of a permanent magnet, Figure 3a. The magnetocaloric effect causes the material in the bed to increase in temperature when it is magnetized. While the bed is in the magnetic field, it experiences a flow of heat transfer fluid from its cold end to its hot end; this flow causes a heat rejection in the hot heat exchanger (Figure 3b) because the temperature of the fluid leaving the hot end is hotter than the ambient temperature. The bed is demagnetized as it rotates out of the permanent magnet (Figure 3c), causing the temperature of the bed to decrease. The regenerator then experiences a flow of heat transfer fluid from its hot end to its cold end while it is out of the magnetic field (Figure 3d), which causes a cooling load to be accepted at the cold heat exchanger because the temperature of the fluid leaving the bed is less than the refrigeration load temperature.
[FIGURE 3 OMITTED]
DEVELOPMENTS IN MAGNETOCALORIC MATERIALS
The properties of the magnetocaloric material that is used in an AMRR system are primarily responsible for the system performance that can be achieved. A review of recently developed materials for room-temperature refrigeration is given by Brueck (2005). Recently, researchers have developed several promising materials with large magnetocaloric effects and tunable Curie temperatures that may be suitable for room-temperature applications (Gschneidner et al. 2005). Magnetocaloric materials generally have nonlinear properties that are highly dependent on temperature; therefore, evaluating the relative performance of one material compared to another is not straightforward. A rigorous comparison of materials would require that the properties be integrated with a detailed model of the AMRR system, and even then, the results would depend on the regenerator geometry, operating temperatures, heat transfer fluid properties, and several other system or operating parameters.
Although no simple set of properties can define the performance of a magnetocaloric material used in an AMRR, the two parameters that provide the most meaningful basis for comparison are the adiabatic temperature change with magnetization ([T.sub.ad]) and the specific entropy change with magnetization ([DELTA][s.sub.M]). Many magnetocaloric materials exhibit magnetic hysteresis, where material properties are dependent on the history of the magnetic field. Hysteresis will reduce the performance of an AMRR system and therefore should also be considered when selecting a magnetocaloric material. The thermal conductivity of the magnetocaloric material also has an important albeit less intuitive impact on the performance of AMRR devices. A material with a large thermal conductivity could lead to a regenerator that is plagued by large axial conduction, which can be a major loss mechanism for AMRRs. However, it is possible to reduce axial conduction losses by placing low conductivity spacers in the regenerator. Conversely, a material with a low conductivity will not interact completely with the heat transfer fluid during each cycle; the diffusive conduction wave that transfers energy between the material and the fluid will travel too slowly, and, therefore, the material in the center of the solid matrix (e.g., at the center of a spherical particle) will not participate thermally in the AMRR cycle. Regenerators characterized by high fluid-to-solid heat transfer coefficients or high operating frequency are particularly susceptible to losses related to temperature gradients that are internal to the solid material (Engelbrecht et al. 2006a). Therefore, materials with either very high or very low thermal conductivity may not be suited for some AMRR applications; the threshold conductivities depend strongly on the particular geometry of the regenerator that is being considered.
The Curie point is the temperature at which a material changes from a ferromagnetic to a paramagnetic state; this temperature is significant because a material will exhibit its greatest magnetocaloric effect near the Curie temperature. There are two types of magnetic phase changes that may occur at the Curie point, first-order magnetic transition (FOMT) and second-order magnetic transition (SOMT). For SOMT materials, the magnetic moments of the material become aligned during the transformation from a ferromagnet to a paramagnet. There is no discontinuous jump in magnetization and no latent heat at the transition associated with the transition. FOMT materials experience a simultaneous ordering of magnetic dipoles and a latent heat associated with the transition. Some FOMT materials experience a change in the crystal sub-lattice associated with the phase change at the Curie point. According to Gschneidner et al. (2005), the temperature change for SOMT materials upon magnetization is almost instantaneous (on the order of nanoseconds). However, for FOMT materials that experience a change in structure, atoms are displaced during the change in crystal structure and therefore the time required to achieve a temperature change when magnetizing some FOMT materials can be many orders of magnitude larger than the time-scale associated with SOMT materials. This time lag between the application or removal of a magnetic field and the associated thermal response of an FOMT material may decrease cycle performance by 30%--50% when FOMT materials are used in an AMRR that operates at frequencies from 1 to 10 Hz. However, according to Russek and Zimm (2006), FOMT materials with large magnetocaloric effects that use less expensive raw materials have the potential to be more cost-effective than SOMT materials such as Gd.
Figure 2 shows that magnetic materials exhibit a large magnetocaloric effect only over a narrow temperature range that is near the Curie temperature ([T.sub.Curie]) of the material. As a result, there is only a small temperature range where an AMR composed of a single magnetic material can maintain its otherwise potentially high performance. In order to maximize the magnetocaloric effect over a large temperature span, a regenerator bed composed of several materials may be constructed with an engineered, spatial variation in its Curie temperature, which is selected in order to match the local, average regenerator temperature. A regenerator constructed of several magnetocaloric materials is referred to as a layered regenerator and AMRR systems that utilize layered regenerators have the potential to achieve higher system performance than single-material AMRR systems. Therefore, researchers are working toward the development of families of material compounds that have similar properties but whose Curie temperature can be shifted by changing the material composition. For example, the Curie temperature of alloys of gadolinium, Gd, and dysprosium, Dy, can be adjusted by varying the fraction of each element (Gschneidner et al. 2005). A thorough review of magnetocaloric materials can be found in Gschneidner et al. (2005). This paper will discuss some of the most promising families of materials that were presented in the paper by Gschneidner et al. as well as more recently developed materials that were not included in that review.
According to Allab et al. (2006), the upper limit of the magnetic field strength that can be achieved using a permanent magnet is approximately 2 Tesla; therefore, the properties of magnetocaloric materials are compared using a magnetic field change from 0 to 2 Tesla. A summary of the magnetic properties of selected materials is presented in Table 1. Note that in Table 1, hysteresis is defined as the temperature shift that is observed as the magnetic field increases as compared to its value as the magnetic field decreases.
Table 1. Summary of Properties of Selected Magnetocaloric Materials Material [T.sub.Curie],K Gd 294 G[d.sub.5]S[i.sub.4-x]G[e.sub.x] x = 0.5 269 LaF[e.sub.13-x]S[i.sub.x][H.sub.y] x = 1.3, y = 1.1 287 L[a.sub.1-w]C[e.sub.w]F[e.sub.13-x]S[i.sub.x][H.sub.y] w = 0.1, x = 1.3, y = 1.6 334 MnA[s.sub.1-x]S[b.sub.x] x = 0 318 x = 0.1 287 M[n.sub.1.1s] F[e.sub.0.9][P.sub.1-x]G[e.sub.x] x = 0.2 206 x = 0.22 280 M[n.sub.1.1]F[e.sub.0.9][P.sub.0.5]A[s.sub.0.5] 282 P[r.sub.13]F[e.sub.87] 288 G[d.sub.1-x]T[b.sub.x]C[o.sub.2] x = 0.6 300 x = 1 230 N[i.sub.2 + x]M[n.sub.1-x]Ga x = 0.16 314 x = 0.18 333 Material Type Gd SOMT G[d.sub.5] S[i.sub.4-x]G[e.sub.x] x = 0.5 FOMT LaF[e.sub.13-x]S[i.sub.x][H.sub.y] x = 1.3, y = 1.1 FOMT L[a.sub.1-w]C[e.sub.w]F[e.sub.13-x]S[i.sub.x][H.sub.y] w = 0.1, x = 1.3, y = 1.6 FOMT MnA[s.sub.1-x]S[b.sub.x] x = 0 FOMT x = 0.1 FOMT M[n.sub.1.1] F[e.sub.0.9][P.sub.1-x]G[e.sub.x] x = 0.2 SOMT x = 0.22 FOMT M[n.sub.1.1]F[e.sub.0.9][P.sub.0.5]A[s.sub.0.5] FOMT P[r.sub.13]F[e.sub.87] SOMT G[d.sub.1-x]T[b.sub.x]C[o.sub.2] x = 0.6 SOMT x = 1 SOMT N[i.sub.2 + x]M[n.sub.1-x]Ga x = 0.16 FOMT x = 0.18 FOMT Material -[DELTA] [s.sub.M] (0-2 Tesla), J/kg.K Gd 5.8 G[d.sub.5] S[i.sub.4-x]G[e.sub.x] x = 0.5 27 LaF[e.sub.13-x]S[i.sub.x][H.sub.y] x = 1.3, y = 1.1 28 L[a.sub.1-w]C[e.sub.w]F[e.sub.13-x]S[i.sub.x][H.sub.y] w = 0.1, x = 1.3, y = 1.6 30 MnA[s.sub.1-x]S[b.sub.x] x = 0 32 x = 0.1 30 M[n.sub.1.1] F[e.sub.0.9][P.sub.1-x]G[e.sub.x] x = 0.2 13 x = 0.22 16 M[n.sub.1.1]F[e.sub.0.9][P.sub.0.5]A[s.sub.0.5] 25 P[r.sub.13]F[e.sub.87] 3 G[d.sub.1-x]T[b.sub.x]C[o.sub.2] x = 0.6 2.5 x = 1 3.5 N[i.sub.2 + x]M[n.sub.1-x]Ga x = 0.16 11 (0-1.8 Tesla) x = 0.18 21 Material [DELTA] [T.sub.ad] (0-2 Tesla), K Gd 5.5 G[d.sub.5] S[i.sub.4-x]G[e.sub.x] x = 0.5 7 LaF[e.sub.13-x]S[i.sub.x][H.sub.y] x = 1.3, y = 1.1 7.1 L[a.sub.1-w]C[e.sub.w]F[e.sub.13-x]S[i.sub.x][H.sub.y] w = 0.1, x = 1.3, y = 1.6 MnA[s.sub.1-x]S[b.sub.x] x = 0 5 x = 0.1 3 (0-1.45 Tesla) M[n.sub.1.1] F[e.sub.0.9][P.sub.1-x]G[e.sub.x] x = 0.2 x = 0.22 M[n.sub.1.1]F[e.sub.0.9][P.sub.0.5]A[s.sub.0.5] P[r.sub.13]F[e.sub.87] G[d.sub.1-x]T[b.sub.x]C[o.sub.2] x = 0.6 x = 1 N[i.sub.2 + x]M[n.sub.1-x]Ga x = 0.16 x = 0.18 Material Hysteresis, K Gd ~0 G[d.sub.5] S[i.sub.4-x]G[e.sub.x] x = 0.5 2 LaF[e.sub.13-x]S[i.sub.x][H.sub.y] x = 1.3, y = 1.1 1 L[a.sub.1-w]C[e.sub.w]F[e.sub.13-x]S[i.sub.x][H.sub.y] w = 0.1, x = 1.3, y = 1.6 MnA[s.sub.1-x]S[b.sub.x] x = 0 6 x = 0.1 1 M[n.sub.1.1] F[e.sub.0.9][P.sub.1-x]G[e.sub.x] x = 0.2 ~0 x = 0.22 11 M[n.sub.1.1]F[e.sub.0.9][P.sub.0.5]A[s.sub.0.5] P[r.sub.13]F[e.sub.87] G[d.sub.1-x]T[b.sub.x]C[o.sub.2] x = 0.6 x = 1 N[i.sub.2 + x]M[n.sub.1-x]Ga x = 0.16 7 x = 0.18 7 Material Reference Gd Dan'kov (1998) G[d.sub.5] S[i.sub.4-x]G[e.sub.x] x = 0.5 Pecharsky et al. (2003) LaF[e.sub.13-x]S[i.sub.x][H.sub.y] x = 1.3, y = 1.1 Fujita et al. (2003) L[a.sub.1-w]C[e.sub.w]F[e.sub.13-x]S[i.sub.x][H.sub.y] w = 0.1, x = 1.3, y = 1.6 Fujita et al. (2004) MnA[s.sub.1-x]S[b.sub.x] x = 0 318 x = 0 Wada and Tanabe (2001) x = 0.1 Wada et al. (2005) M[n.sub.1.1] F[e.sub.0.9][P.sub.1-x]G[e.sub.x] x = 0.2 Yan et al. (2005) x = 0.22 Dagula et al. (2005) M[n.sub.1.1]F[e.sub.0.9][P.sub.0.5]A[s.sub.0.5] Tegus et al. (2004) P[r.sub.13]F[e.sub.87] Pawlik et al. (2006) G[d.sub.1-x]T[b.sub.x]C[o.sub.2] x = 0.6 Zhou et al (2006) x = 1 Zhou et al (2006) N[i.sub.2 + x]M[n.sub.1-x]Ga x = 0.16 Cherechukin et al.(2004) x = 0.18 Cherechukin et al.(2004)
Gadolinium and Its Alloys
Gadolinium, Gd, is an SOMT material with a Curie temperature of approximately 293 K. It is the only pure substance with a Curie point near room temperature and exhibits a significant magnetocaloric effect over a large temperature span. Dan'kov et al. (1998) studied the magnetocaloric properties of Gd and found the maximum [DELTA][T.sub.ad] to be approximately 5.8 K when magnetized from 0 to 2 Tesla using a direct measurement of [DELTA][T.sub.ad]. The magnetic hysteresis exhibited by gadolinium is quite low and Dan'kov et al. reported that there was no detectable hysteresis in single gadolinium crystals. The thermal conductivity of Gd near room temperature is approximately 10 W/m.K (Fujieda et al. 2004). Because gadolinium has a relatively large magnetocaloric effect and low hysteresis, it has been used in many prototype room temperature AMRR systems (Yu et al. 2003). Gadolinium can corrode in the presence of water at room temperature, which could affect the long-term performance and durability of an AMRR system. However, Zhang et al. (2005) found that by adding NaOH to the water, this corrosion problem can be eliminated. Practical AMRR systems using gadolinium will likely require some type of corrosion inhibitor in the heat transfer fluid.
Gadolinium can be alloyed with terbium (Tb) (Gschneidner and Pecharsky 2000), dysprosium (Dy) (Dai et al. 2000), or erbium (Er) (Nikitin et al. 1985) in order to lower the Curie temperature. Canepa et al. (2002) report that palladium (Pd) can be added to Gd to form [Gd.sub.7][Pd.sub.3], which has a higher Curie point than pure Gd. All these Gd alloys exhibit magnetocaloric properties that are similar to pure Gd, and these families of Gd alloys can be used to construct a layered regenerator bed.
Alloys composed of gadolinium, silicon, and germanium exhibit a magnetocaloric effect that is significantly larger than that of gadolinium and have a phase transition temperature that is near room temperature (Pecharsky and Gschneidner 1997a). The Curie temperature of the material can be adjusted by varying the fraction of silicon and compounds with a wide range of Curie temperatures can be synthesized (Pecharsky and Gschneidner 1997b). Unlike gadolinium, most G[d.sub.5]S[i.sub.4-x][Ge.sub.x] compounds are FOMT materials, and the entropy change with magnetization is larger than that of gadolinium but is also much sharper, occurring over a narrower temperature range. The FOMT involves a change in crystal symmetry, with a magnetic hysteresis that is larger than that of gadolinium, with a value of 2 K reported for G[d.sub.5]S[i.sub.2]G[e.sub.2] (Pecharsky and Gschneidner 1997b). The thermal conductivity of G[d.sub.5]S[i.sub.2]G[e.sub.2] was experimentally determined to be approximately 5-7 W/m.K near room temperature (Fujieda et al. 2004). The hysteresis can be greatly reduced by alloying with other elements, but the material then becomes an SOMT (Pecharsky and Gschneidner 1997c; Shull et al. 2006).
Pecharsky et al. (2003) found that by using high-purity starting components and a different heat treatment than was used in the previous work by Pecharsky and Gschneidner (1997a), the entropy change with magnetization and the adiabatic temperature change for G[d.sub.5]S[i.sub.2]G[e.sub.2] could both be increased by more than 50%. The entropy change, [DELTA][s.sub.M], of this optimally prepared material was approximately -27 J/kg.K and the adiabatic temperature change, [DELTA][T.sub.[ad]], was approximately 7 K when the material was magnetized from 0 to 2 Tesla. The adiabatic temperature change was determined from heat capacity and magnetization measurements and was not measured directly. G[d.sub.5] (S[i.sub.[1-x]]G[e.sub.x])4 materials have the potential to be high-performance AMRR refrigerants because they possess a relatively high entropy change with magnetization and a large adiabatic temperature change. Families of these materials with similar magnetic properties can be fabricated with large spans of Curie temperature.
Alloys of lanthanum, iron, silicon, and hydrogen undergo a first-order magnetic phase transition and exhibit a magnetocaloric effect that is larger in magnitude than gadolinium but is much sharper (i.e., it occurs over a narrower temperature span), as discussed by Fujieda et al. (2002). The properties of the material can be adjusted by substituting iron for silicon (Gschneidner et al. 2005) or by adding hydrogen (Fujita et al. 2003). The Curie temperature of this family of materials has been reported from 195 K to 336 K, depending on composition. La(F[e.sub.[11.7]]S[i.sub.[1.3]])[H.sub.[1.1]] has a Curie temperature of 287 K and experiences an entropy change of -28 J/kg.K and an adiabatic temperature rise of 7.1 K when magnetized from 0 to 2 Tesla. The adiabatic temperature rise was measured indirectly using heat capacity and magnetization data. The hysteresis for LaF[e.sub.[11.44]]S[i.sub.[1.56]] is approximately 1 K according to Fujita et al. (2003); however, the hysteresis can vary widely with material composition. The thermal conductivity of these materials is approximately 10 W/m.K near room temperature (Fujieda et al. 2004). La is one of the most common rare earth elements, with a cost that is significantly less than Gd, making these materials potentially more desirable economically than Gd.
The performance of this family of materials can be enhanced by substituting other elements for lanthanum. By substituting cerium (Ce) for lanthanum, Fujieda et al. (2004) were able to increase the entropy change with magnetization to -30 J/kg.K for a compound with 10% cerium substitution. Fujieda et al. (2006) found that substituting praseodymium (Pr) for lanthanum increases the entropy change and adiabatic temperature change when magnetized from 0 to 5 Tesla by more than 30% compared to a similar material with no lanthanum substitution. No data were found for a 2 Tesla magnetic field change; however, the material with praseodymium substituted for lanthanum likely has a similar advantage over the unsubstituted material at 2 Tesla. The hysteresis is not affected by substituting praseodymium for lanthanum.
The recently developed Mn[As.sub.[1-x]][Sb.sub.x] compounds are FOMT materials that may be suitable for magnetic refrigeration systems. MnAs has a Curie temperature of approximately 318 K, an entropy change with magnetization, [DELTA][s.sub.M], of approximately -31 J/kg.K, and an adiabatic temperature change, [DELTA][T.sub.[ad]], of approximately 5 K when magnetized from 0 to 2 Tesla (Wada and Tanabe 2001). The adiabatic temperature was measured indirectly using heat capacity measurements. MnAs has a relatively large hysteresis of 5 K and low thermal conductivity of approximately 2 W/m.K near room temperature (Fujieda et al. 2004). The Curie temperature of this alloy can be adjusted between 230 K and 318 K by substituting antimony (Sb) for arsenic (As). When the fraction of Sb (x) is greater than or equal to 0.05, the thermal hysteresis becomes quite small while the magnetocaloric effect remains approximately unchanged (Wada et al. 2005). For the material containing Sb,[DELTA][T.sub.[ad]] was determined to be 3 K when magnetized from 0 to 1.45 Tesla using a direct measurement of adiabatic temperature change. These materials are attractive as magnetic refrigerants because they have a large entropy change with magnetization, the Curie temperature can be adjusted over a large temperature range, and the hysteresis becomes relatively small with the addition of antimony. However, the adiabatic temperature change is relatively low for this family of materials and the thermal conductivity is significantly lower than that of gadolinium and the other magnetocaloric materials that have been discussed; these properties may decrease performance under some AMRR operating conditions.
The properties of this family of materials can be modified with the addition of elements As, Mn, Ge, cobalt (Co), and chromium (Cr) (Tegus et al. 2004). The Curie temperature of [Mn.sub.1.1][Fe.sub.0.9][P.sub.1-x][Ge.sub.x] can be adjusted between 250 K and 380 K by varying x, the fraction of germanium (Ge) in the compound (Dagula et al. 2005). Yan et al. (2006) report that the properties of this alloy are strongly dependent on the material processing technique that is used. The Curie temperature of a melt-spun alloy was found to be 18 K higher than that of an annealed bulk alloy with the same composition. The melt-spun materials also exhibited higher entropy change with magnetization and lower hysteresis. This family of materials is generally FOMT; however, when x = 0.2, the material becomes SOMT with no detectable hysteresis. The Curie temperature for the x = 0.2 alloy is approximately 206 K, and, therefore, it is not well suited for space-cooling applications but may be useful for some low-temperature refrigeration applications. As germanium is substituted for phosphorus (P), the thermal hysteresis increases and reaches a value of approximately 8 K for compounds with x > 0.2. The entropy change with magnetization is approximately -16 J/kg K for a 0 to 2 Tesla applied field change for materials that have a Curie temperature near room temperature. The Curie temperature of [Mn.sub.1+x][Fe.sub.1-x][P.sub.1-y][As.sub.y] can be adjusted by varying the fraction of As. The maximum entropy change for these materials from Brueck et al. (2005) is -20 J/kg.K for [Mn.sub.1.1][Fe.sub.0.9][P.sub.0.47][As.sub.0.53], which has a Curie temperature of 289 K. Tegus et al. (2004) reported an entropy change with magnetization of approximately -25 J/kg.K for [Mn.sub.1.1][Fe.sub.0.9][P.sub.0.5][As.sub.0.5], which has a Curie temperature of 282 K. These materials have a substantial magnetocaloric effect, and the Curie temperature can be adjusted over a wide temperature range, which makes them competitive as possible magnetic refrigerants. The thermal hysteresis for these materials is larger than for other materials mentioned above, which could reduce performance in an AMRR system.
Compounds of [Ni.sub.2+x][Mn.sub.1-x]Ga have Curie temperatures that lie between 315 and 380 K and have a relatively large magnetocaloric effect. Cherechukin et al. (2004) studied these materials and found that the entropy change of magnetization when magnetized from 0 to 1.8 Tesla is -20.7 J/kg.K for [Ni.sub.2.18][Mn.sub.0.82]Ga. These materials have a relatively high hysteresis of 7 K (Gschneidner et al. 2005), which makes them less desirable than other materials with large magnetocaloric effects. However, new fabrication processes may reduce this hysteresis, making these materials well suited for coolers operating at higher temperatures, for example between 300 and 350 K.
Several other materials have been developed recently, but none seem as promising as the materials discussed above. Dinesen et al. (2005) studied the magnetic properties of [La.sub.0.67][Ca.sub.0.33][Sr.sub.x][MnO.sub.3[+ or -][delta]] and found that the adiabatic temperature change of these materials is significantly lower than that of gadolinium. For a Sr fraction of 0.55, [DELTA][T.sub.ad] is 1 K and [DELTA].[s.sub.M] is -2.8 J/kg.K when magnetized from 0 to 1.2 Tesla. Pawlik et al. (2006) studied alloys of praseodymium and iron (Fe) and found that the maximum magnetocaloric effect occurred in [Pr.sub.13][Fe.sub.87], with an entropy change of magnetization of approximately -3 J/kg.K, which is less than that of Gd. Compounds of Gd, Tb, and Co were studied by Zhou et al. (2006) and the maximum entropy change with magnetization was found to be -3.6 J/kg.K, which is also significantly lower than that of pure Gd.
AMRR PROTOTYPE SYSTEMS
Different mechanical realizations of the AMRR cycle are possible, and several types of prototype systems have been constructed recently. For example, the magnetic material may be stationary and the field varied by controlling the current in a solenoid; however, this configuration is currently only practical at very low temperatures or where superconductors can be used to efficiently handle the large currents that are required to generate useful magnetic fields. Practical AMRR systems used for residential and other smaller-scale applications will likely use a permanent magnet to generate a magnetic field because the power required by the cryogenic equipment necessary to maintain the superconducting temperature of a solenoid magnet can greatly exceed the cooling power of small-to medium-scale AMRR devices (Zimm et al. 2006). However, for large-scale systems, the increased performance that is possible with superconducting magnets may offset the power required to maintain the magnet at a cryogenic temperature. Systems using permanent magnets can achieve variations in the applied field by physically moving the magnetic regenerator relative to the magnetic field, either linearly in a reciprocating device or rotationally in a rotary device (e.g., Figure 3); a schematic of a reciprocating AMRR system is shown in Figure 4.
Several AMRR prototypes have been built and their measured performance has been recently published. Some of these prototypes have implemented layered regenerator beds in an attempt to increase performance, and many of these newer systems are using the more practical option of permanent magnets rather than superconducting solenoid magnets to generate the magnetic field. Yu et al. (2003) provide a summary of the prototype systems and their associated performance that covers the period up to 2003; only the most significant systems that were discussed in that paper are cited here. More recent results for new prototype systems and new system configurations are summarized in Table 2 and discussed below. Examination of Table 2 shows that the performance of AMRR systems is highly dependent on the operating conditions used for the tests and materials used to fabricate the regenerator. In every case, the maximum value of the cooling power ([Q.sub.C]) that can be produced by a system will occur as the temperature span of the system ([DELTA]T) approaches zero, and the maximum temperature span will occur when no refrigeration load is produced. One must, therefore, pay close attention to the system operating conditions when evaluating and comparing the performance of the prototype AMRR systems presented in Table 2 and discussed in this section.
Table 2. Summary of Experimental Performance of Recently Built AMRR Systems System Config. [[mu].sub.o] # [H.sub.max], Beds Tesla Brown Device recip. 7(E) 1 US Navy recip 7(E) 1 2(S) University of recip. 2(S) 2 Victoria 2(S) 2(S) Chubu 4(S) Electric/ recip. 2(S) 2 Toshiba Astronautics recip. 5(S) 2 Recip. Astronautics rotary 1.5 6 Rotary (P) Grenoble recip. 0.8 1 (P) Nanjing recip. 1.4 2 University (P) Tokyo Institute of Technology/Chubu rotary 0.77 4 (P) Xi'an Jiaotong recip. 2.18 1 University (S) System Regen. Freq., Regen. [Q.sub.c], Vol., Hz Material W [cm.sup.3] Brown Device not Gd 0 spec. US Navy 173 0.01 Gd 0 University of 74 1 Gd, GdTb 0 Victoria &GdEr 49 0.8 Gd 0 25 0.6 7 Chubu Electric/ 484 0.167 Gd 100 Toshiba 40 Astronautics ~600 0.167 Gd 100 Recip. 600 Astronautics 33 4 Gd 15 Rotary Gd & 27 GdEr 0 Grenoble 32 0.42 Gd 9 Gd 0 Nanjing ~200 <0.25 Gd5Si2Ge2 0 University Hz 0 GdSiGeGa 40 Tokyo Institute 0.42 60 of layered Technology/Chubu ~200 0.55 Gdl-xDyx 14 Xi'an Jiaotong ~200 ~0.1 Gd 19 University Hz Gd5Si2Ge2 10 System T,K Regenerator Reference Geometry Brown Device 47 parallel Brown (1976) plates (1.0 mm) US Navy 40 embossed Green (1986) ribbon (0.2 mm) University of 50 crashed Rowe et al. Victoria part. (2006) (0.25-0.65 mm) 15.5 spheres (0.2 Rowe et al. 14 mm) (2004) Chubu 26 spheres Hirano et al. Electric/ 24 (0.3 mm) (2002) Toshiba Astronautics 38 spheres Zimmet al Recip. 0 (0.15-0.3 (1998) mm) Astronautics 14 spheres Zimmet al. Rotary (0.43-0.5 (2006) 14 mm) spheres 25 (0.25-0.355 mm) Grenoble 4 parallel Clot et al. 23 plates (1.0 (2003) mm) Nanjing 10 spheres (0.2 Luetal.(2005) University 25 mm) 5 Tokyo Institute 0 spheres (0.6 Okamuraet al. of mm) (2006) Technology/Chubu 4 Xi'an Jiaotong 4 spheres (0.3 Yu et al. University 3 mm) (2006) irregular part (0.3-0.75 mm) a [[mu].sub.0][H.sub.[max]]: S, superconducting magnet; E, electromagnet; P, permanent magnet.
University of Victoria Prototype
The University of Victoria AMRR consists of two regenerator beds that are moved linearly through a magnetic field that is generated by a stationary superconducting solenoid magnet (Rowe et al. 2004); a schematic of the system is shown in Figure 4. The maximum magnetic field generated by the magnet is 2 Tesla. The regenerator is constructed by stacking from 1 to 3 "pucks" of magnetocaloric material. Each puck is 2.5 cm in diameter and 2.5 cm long and can be made of a different material with a different geometry. The heat transfer fluid is helium at 10 bar and flow is controlled by a displacer. The cooling load is controlled by two 25 W finned heaters that are mounted between the two regenerator beds. The device can operate from 0.2 to 1 Hz; the maximum frequency is constrained by the inertial forces.
[FIGURE 4 OMITTED
The prototype has been tested with several layered regenerator beds over a range of operating frequencies. The performance of a layered bed was compared directly to a single-material bed frequencies. The performance of a layered bed was compared directly to a single-material bed by first running the system with 2 Gd pucks and then with one Gd puck and one [Gd.sub.0.74][Tb.sub.0.26] puck. The layered bed produced a no-load temperature span of approximately 19 K, while the single-material bed produced a span of approximately 15.5 K (Rowe et al. 2004). Using a single "puck" of Gd, the prototype produced 7 W of cooling over a temperature range of 288--302 K. Rowe and Tura (2006) constructed a bed of three "pucks" consisting of Gd, [Gd.sub.0.74][Tb.sub.0.26], and [Gd.sub.0.85][Er.sub.0.15] in order of decreasing Curie temperature. Operating at a frequency of 1 Hz, a no-load temperature span of 51 K was achieved, showing that layered beds allow AMRR systems to operate over a substantially larger temperature span than can be achieved using a single-material regenerator.
Chubu Electric and Toshiba Device
The Chubu Electric/Toshiba AMRR has two regenerator beds that are moved linearly in the presence of a 4 Tesla magnetic field that is generated by a superconducting solenoid. Each regenerator has a diameter of 6.2 cm and a length of 8 cm and is composed of packed spheres of Gd. The heat transfer fluid is a mixture of water and ethanol. Fluid flow is controlled by two valves and a variable-speed pump. Hirano et al. (2002) report a cooling power of 100 W with a COP of 5.6 when the system operates between 302 K and 276 K. However, the reported COP is somewhat misleading because it does not include any motor or pump inefficiencies or the power that is required to cool the superconducting solenoid. When the maximum magnetic field is decreased to 2 Tesla, the cooling power drops to approximately 40 W between 298 and 274 K, showing that the strength of the magnetic field can significantly affect AMRR performance.
Astronautics Reciprocating and Rotary Prototypes
Astronautics' first near-room-temperature AMRR prototype was a reciprocating device with a 5 Tesla magnetic field generated by a superconducting solenoid. Two regenerators were made of packed Gd spheres of 0.15-0.30 mm diameter. The regenerator material occupied a total volume of approximately 600 [cm.sup.3]. The maximum cooling power for this device is 600 W, which is obtained as the temperature span approaches 0, and a cooling power of 100 W was achieved with a temperature span of 38 K (Zimm et al. 1998). Although the original device produced a relatively large cooling power over a large temperature span, the device itself was quite large and used a superconducting magnet and therefore would not be practical as a commercial product (Zimm et al. 2006).
Recently Astronautics has built a rotary device that is more compact and uses a more practical 1.5 Tesla [Nd.sub.2][Fe.sub.14]B permanent magnet. The regenerator is divided into six separate beds that rotate through the field of the permanent magnet, as shown in Figure 3. The heat transfer fluid is water based and the fluid flow rate is controlled using a variable-speed pump and two rotary valves. The plumbing of the system is set up so that the pump runs continuously, and flow through the external piping and heat exchangers is unidirectional while the flow through each individual bed reverses direction during operation. The rotary configuration allows the device to operate at frequencies up to 4 Hz (240 RPM), which is higher than any reciprocating AMRR built to date.
Zimm et al. (2006) studied the performance of a layered bed of Gd and [Gd.sub.0.94][Er.sub.0.06] and a single-material bed of a LaFeSiH material using the rotary device. They found that the layered bed is capable of producing a larger cooling load than a similar bed made of Gd when the temperature span is large. The rotary device achieved a maximum no-load temperature span of approximately 25 K with the layered bed and a no-load temperature span of less than 19 K with a single-material Gd bed. The cooling power over a 14 K span for the layered bed was 80% higher (27 W) than the 15 W cooling load that was produced by the single-material regenerator. However, the cooling power is greater for the single-material Gd bed as the temperature span approaches zero; the layered bed was capable of producing 41 W of cooling at zero temperature span while the Gd bed produced 44 W of cooling under the same condition. This behavior was expected since the magnetocaloric effect of Gd is greater than that of [Gd.sub.0.94][Er.sub.0.06] at room temperature. A regenerator made of irregular particles of La([Fe.sub.11.44][Si.sub.1.56c)[H.sub.1.0] with sizes in the range of 0.25-0.5 mm was also studied using the rotary device. The LaFeSiH material produced a higher cooling load than either the Gd or the Gd and GdEr regenerator beds at zero temperature span. However, the LaFeSiH material produced lower cooling power when the temperature span was increased. Based on this result, Zimm et al. concluded that LaFeSiH materials are promising materials for AMRR systems but that a layered bed would be necessary if these alloys are to achieve high performance over a useful temperature span.
Nanjing University built a reciprocating device consisting of two regenerator beds that are moved linearly into and out of the magnetic field generated by a stationary, 1.4 Tesla permanent magnet. The regenerators fit within the 3 cm diameter bore of the magnet. Lu et al. (2005) have operated the system with Gd regenerators as well as several recently developed, more advanced magnetocaloric materials. The maximum reported cooling power for this system is 40 W at a temperature span of 5 K when using a [Gd.sub.5][Si.sub.1.895][Ge.sub.1.89][Ga.sub.0.03] regenerator. The no-load temperature span for this device is 23 K for a single-material Gd regenerator, 25 K for a [Gd.sub.5][Si.sub.1.895] regenerator, and 10 K for a [Gd.sub.5][Si.sub.2][Ge.sub.2] regenerator. It is not clear why the [Gd.sub.5][Si.sub.2][Ge.sub.2] regenerator achieved a lower temperature span than the pure Gd regenerator. However, Shull et al. (2006) found that adding small amounts of materials such as gallium (Ga) to [Gd.sub.5][Si.sub.4-x][Ge.sub.x] materials changes the magnetic characteristics from FOMT to SOMT, increases the Curie temperature of the material, and greatly reduces hysteresis. The reduced hysteresis and increased Curie temperature of the [Gd.sub.5][Si.sub.1.895][Ge.sub.1.89][Ga.sub.0.03] material may account for the dramatic improvement in the no-load temperature span, which is more than double that of the compound without Ga.
Tokyo Institute of Technology and Chubu Rotary Device
The Chubu system is a rotary device that is similar to the Astronautics device except that the magnet rotates and the regenerator is stationary. The regenerator consists of four 0.2 mm packed sphere regenerator beds; each bed consists of four layers of different materials. The layers are, from the cold end to the hot end, [Gd.sub.0.91][Y.sub.0.09], [Gd.sub.0.84][Dy.sub.0.16], [Gd.sub.0.87][Dy.sub.0.13], and [Gd.sub.0.89][Dy.sub.0.11]. During operation, the 0.77 Tesla neodymium permanent magnet rotates over a regenerator bed and then stops while fluid flows though the bed. The magnet then rotates 90[degrees] to the next regenerator bed and the process is continued in this manner. The heat transfer fluid is water, and its flow is controlled by a rotary valve and variable-speed pump. The system produces a maximum cooling power of 60 W at a frequency of 0.42 Hz and a fluid flow rate of 4 L/min as the temperature span approaches 0 K (Okamura et al. 2006). For a temperature span of 4 K, a maximum cooling power of 14 W was achieved at a frequency of 0.55 Hz and a fluid flow rate of 3 L/min.
Clot et al. (2003) built a reciprocating AMRR using 1 mm parallel plates of Gd that are separated by a 0.15 mm gap to allow fluid flow. The magnetic field is provided by a 0.8 Tesla permanent magnet, and the system operates at 0.42 Hz. A cooling load of 8.8 W is obtained for a temperature span from 298.5 to 302.5 K with a COP of 2.2; the COP includes all power inputs to the system.
Xi'an Jiaotong University built a reciprocating AMRR with a single regenerator bed that uses a 2.18 Tesla electromagnet (Yu et al. 2006). The dimensions of the bed were not reported, but the bed holds 930 g of Gd particles. Assuming a porosity of 0.4, the volume of the regenerator is approximately 200 [cm.sup.3]. Using Gd spheres as the refrigerant, this device produced up to 18.7 W of cooling between 294.5 and 291.4 K with a fluid flow rate of 3.5 L/min. The same device was operated using a regenerator made of irregular particles of [Gd.sub.5][Si.sub.2][Ge.sub.2], and the cooling power dropped to a maximum value of 10.3 W between the temperatures of 300.1 and 297.1 K with a fluid flow rate of 3.6 L/min. The reason for the lower cooling power measured with the [Gd.sub.5][Si.sub.2][Ge.sub.2] material is not known; however, it is likely because the system is not operating near the Curie point of the material.
Vasile and Muller (2006) describe a prototype using a rotating permanent magnet and stationary regenerator. The regenerator is composed of a series of parallel plate regenerator "inserts" that are thermally isolated from each other and can be made of different magnetocaloric materials to construct a layered regenerator. However, there is currently little published performance data for this prototype.
PRACTICAL AMRR ISSUES
Magnetic refrigeration is not a mature field and there will be issues that must be overcome before AMRR devices become practical for commercial applications. Recent prototypes have addressed some of the practical issues associated with the previous experiments. For example, some recent systems have demonstrated relatively large cooling powers using permanent magnets; this eliminates the need for a cryocooler and superconducting solenoids. The maximum operating frequency of AMRR systems has been increased by using rotary rather than reciprocating configurations; the increase in the operating frequency allows smaller and therefore more economically viable beds and magnets to be used. Despite these advances, there are still several practical issues that should be considered.
Prototype AMRR systems are currently much larger than the equivalent vapor compression systems that they would replace. For example, Lu et al. (2005) report using a permanent magnet with an outer diameter of 14 cm and a length of 20 cm for a system that provides a maximum cooling power of 40 W. The system, including the magnet, regenerator, pumps, plumbing, and electric motor, is significantly larger than the compressor and electric motor in a vapor compression system; note that the AMRR system would require heat exchangers that are approximately the same size as the evaporator and condenser in a vapor compression system.
As AMRR systems become more efficient through material selection and regenerator design, their size will decrease; however, even a well-designed and advanced AMRR system using an SOMT material will likely be larger than an equivalent vapor compression system. In order for an AMRR with a packed sphere regenerator of Gd to produce 8.8 kW cooling at an efficiency that is competitive with vapor compression, Engelbrecht et al. (2006b) have shown that a total regenerator volume of approximately 4 L will be required. This regenerator volume does not include the volume associated with the magnet, the pump, the valves, or the regenerator housing. Therefore, the total volume of the AMRR system will likely be significantly higher than a vapor compression system that is sized to meet the same load. However, Engelbrecht (2005) also showed that the regenerator volume can be reduced substantially without sacrificing performance if a layered bed composed of low-hysteresis FOMT material compounds is used in conjunction with more sophisticated regenerator geometries. If the material operates in the region of its FOMT, the high latent heat of the materials offers important advantages. First, the latent heat will minimize the reduction in effective [DELTA][T.sub.ad] by the pore fluid volume, allowing use of a more porous bed with lower pressure drop and lower longitudinal conduction, reducing losses. Second, the large latent heat will allow higher specific fluid flow rates for a given AMRR frequency, allowing the use of a much smaller regenerator bed and correspondingly smaller magnet.
High Fluid Mass Flow Rate
Unlike vapor compression systems, the fluid in an AMRR cycle does not undergo a phase change. As shown in Table 1, the adiabatic temperature rise of current magnetocaloric materials is less than 8 K for a reasonable magnetic field swing, and the temperature change of the heat transfer fluid will be lower than this value due to heat transfer losses and the specific heat of the fluid. The heat transfer from the fluid to the load is the product of the mass flow rate, the temperature change of the fluid, and the specific heat of the fluid. Therefore, for a given cooling capacity, the fluid mass flow rate in an AMRR system must be substantially larger than the flow rate of refrigerant in a vapor compression cycle.
Zimm et al. (2006) reported an AMRR cooling power of 30 W over an 8 K temperature span, which required a fluid flow rate of approximately 0.7 kg/min; this equates to a cooling power to fluid mass flow rate ratio of approximately 43 W.min/kg. Engelbrecht et al. (2006b) predict that the flow rate of water in a packed sphere AMRR that produces 8.8 kW of cooling is approximately 84 kg/min, which leads to a cooling power to fluid flow ratio of approximately 105 W.min/kg. These two data points suggest that the cooling power to mass flow rate ratio of an AMRR system will be in the range of 25--150 W.min/kg depending on the cycle design and magnetocaloric material that is used.
For a vapor compression system, the cooling power to fluid mass flow rate ratio will be much higher due to the latent heat of vaporization of the refrigerant. For example, the DOE/ORNL heat pump model (Rice 2006) predicts that a vapor compression system sized to produce 8.8 kW of cooling will require a refrigerant flow rate of 3.3 kg/min; this corresponds to a cooling power to fluid mass flow rate ratio of approximately 2700 W.min/kg. For an equal fluid flow rate, the predicted cooling power of a vapor compression system is approximately 25 times higher than the predicted value of a well-designed AMRR system and approximately 60 times the experimental value reported by Zimm et al. (2006). This high fluid mass flow rate must be considered carefully in the design of an AMRR system to prevent excessive pumping losses in the heat exchangers and connecting piping. Proper care must be taken in the heat exchanger circuiting and the design of the connecting piping and valves associated with an AMRR systems in order to limit parasitic pumping losses. The single-phase fluid of the AMRR can allow the use of improved heat exchanger designs that may make up for the higher fluid flow rate in some applications. In air-conditioning applications, the heat exchange is dominated by the air-side heat exchange resistance, so the increased fluid flow rate may not have much effect on heat exchanger size or cost. Also, because the pressure in AMRR systems is much lower than vapor cycle systems, use of plastic piping becomes possible, which may reduce connecting piping and installation costs.
Processing of magnetocaloric materials may be an issue for commercial AMRR devices. Most materials that have been developed recently have only been produced in laboratory scale quantities, and many of these require lengthy heat treatments or high-purity starting materials in order to achieve optimum magnetocaloric properties. For example, arc-melted samples of LaF[e.sub.x]S[i.sub.1-x][H.sub.y] were annealed for ten days at 1323 K (Fujita et al. 2003) and MnA[s.sub.1-x][Sb.sub.x] compounds were heat-treated for seven days (Wada et al. 2005) in order to achieve desired results. Many magnetocaloric materials are sensitive to the purity of the starting elements that are used to synthesize the material (Tishin 2005), which may make the cost of raw materials very high. Pecharsky et al. (2003) reported significant increases in entropy change and adiabatic temperature change with magnetization when using 99.9% pure Gd instead of Gd with a purity of 95%--98%. The requirement for materials of high purity, long heat treatments, and expensive raw materials may make the commercial production of some types of magnetocaloric materials impractical.
Recent prototype AMRR systems have proved that layered regenerator beds can produce larger cooling capacity than single-material regenerators when the regenerator materials are properly chosen in order to match the working temperature range. Other prototypes have shown that AMRR systems can provide cooling loads over a relatively large temperature span using practical permanent magnets rather than superconducting solenoid magnets. New materials with high magnetocaloric effects continue to be developed; however, in practice, Gd and its alloys continue to produce the largest cooling power and largest no-load temperature span. FOMT materials generally exhibit higher entropy change of magnetization and adiabatic temperature change than SOMT materials; however, to date the only FOMT material that has been experimentally shown to outperform Gd and its alloys is a [LaFe.sub.x]S[i.sub.1-x][H.sub.y] compound (Zimm et al. 2006). The lower experimental performance of some FOMT materials may be related to the higher hysteresis and longer time required for the magnetic phase change associated with FOMT materials as compared to SOMT materials. Although there have been some promising advances in the field of magnetic refrigeration, there are still many practical issues that must be addressed before AMRR systems become competitive with vapor compression systems for residential applications.
The technical assistance of Andrew Rowe of the University of Victoria is greatly appreciated. This work was funded by an ASHRAE Grant-In-Aid and the University of Wisconsin-Madison Graduate School.
P = pressure, Pa
[Q.sub.C] = cooling power at cold end, W
S = entropy, J/K
[DELTA][s.sub.M] = specific entropy change with magnetization, J/kg.K
T = temperature, K
[DELTA][T.sub.ad] = adiabatic temperature change with magnetization, K
[T.sub.Curie] = Curie temperature, K
U = internal energy, J
V = volume, [m.sup.3]
[[mu].sub.0]H = applied field, Tesla
C = cold or refrigeration temperature
H = hot or heat rejection temperature
Allab, F., A. Kedous-Lebouc, J.P. Yonnet, and J.M. Fournier. 2006. A magnetic field source system for magnetic refrigeration and its interaction with magnetocaloric material. International Journal of Refrigeration 29(8):1340-47.
Brown, G.V. 1976. Magnetic heat pumping near room temperature. Journal of Applied Physics 47:3673-80.
Brueck, E. 2005. Developments in magnetocaloric refrigeration. Journal of Physics D 38:R381-R391.
Brueck, E., M. Ilyn, A.M. Tishin, and O. Tegus. 2005. Magnetocaloric effects in MnFe[P.sub.1-x][As.sub.x]-based compounds. Journal of Magnetism and Magnetic Materials 290-91:8-13.
Canepa, F., S. Cirafici, M. Napoletano, and F. Merlo. 2002. Magnetocaloric properties of [Gd.sub.7]P[d.sub.3] and related intermetallic compounds. IEEE Transactions on Magnetics 38(5):3249-51.
Clot, P., D. Viallet, F. Allab, A. Kedous-Lebouc, J.M. Fournier, and J.P. Yonnet. 2003. A magnet-based device for active magnetic regenerative refrigeration. IEEE Transactions on Magnetics 39(5):3349-51.
Dagula, W., O. Tegus, B. Fuquan, L. Zhang, P.Z. Si, M. Zhang, W.S. Zhang, E. Brueck, F.R. de Boer, and K.H. Buschow. 2005. Magnetic-entropy change in [Mn.sub.1.1][Fe.sub.0.9][P.sub.1-x][Ge.sub.x] compounds. IEEE Transactions of Magnetics 41(10):2778-80.
Dai, W., B.G. Shen, D.X. Li, and Z.X. Gao. 2000. New magnetic refrigeration materials for temperature from 165 K to 235 K. Journal of Alloys and Compounds 311(1):22-25.
Dan'kov, S.Y., A.M. Tishin, V.K. Pecharsky, and K.A. Gschneidner. 1998. Magnetic phase transitions and the magnetothermal properties of gadolinium. Physical Review B 57(6):3478-90.
Dinesen, A.R., L. Linderoth, and S. Morup. 2005. Direct and indirect measurement of the magnetocaloric effect in [La.sub.0.67][Ca.sub.0.33-x][Sr.sub.x][[MnO.sub.3[+ or-][delta]. Journal of Physics: Condensed Matter 17:6257-69.
Engelbrecht, K. 2005. A numerical model of an active magnetic regenerator refrigeration system. Master's thesis, University of Wisconsin-Madison, Madison, WI.
Engelbrecht, K.L., G.F. Nellis, and S.A. Klein. 2006a. The effect of internal temperature gradients on regenerator matrix performance. Journal of Heat Transfer 128(10):1060-69.
Engelbrecht, K.L., G.F. Nellis, and S.A. Klein. 2006b. Predicting the performance of an active magnetic regenerator refrigerator used for space cooling and refrigeration. HVAC&R Research 12(4):1077-95.
Fujieda, S., A. Fujita, and K. Fukamichi. 2002. Large magnetocaloric effect in La([Fe.sub.x][Si.sub.1-x])13 itinerant-electron metamagnetic compounds. Applied Physics Letters 81(7):1276-78.
Fujieda, S., A. Fujita, and K. Fukamichi. 2004a. Enhancements of magnetocaloric effects in La([Fe.sub.0.90][Si.sub.0.10][).sub.13] and its hydride by partial substitution of Ce for La. Materials Transactions 45(11):3228-31.
Fujieda, S., A. Fujita, and K. Fukamichi. 2006. Control of large magnetocaloric effects in [La.sub.1-z][Pr.sub.z]([Fe.sub.x][Si.sub.1-x][).sub.13] Hy magnetic refrigerants working around room temperature. International Journal of Refrigeration 29(8):1302-06.
Fujieda, S., Y. Hasegawa, A. Fujita, and K. Fukamichi. 2004b. Thermal transport properties of magnetic refrigerants La([Fe.sub.x][Si.sub.1-x][).sub.13] and their hydrides, and [Gd.sub.5][Si.sub.2][Ge.sub.2] and MnAs. Journal of Applied Physics 95(5):2429-31.
Fujita, A, S. Fujieda, Y. Hasegawa, and K. Fukamichi. 2003. Itinerant-electron metamagnetic transition and large magnetocaloric effects in La([Fe.sub.x][Si.sub.1-x][).sub.13], compounds and their hydrides. Physical Review B 67(104416):1-11.
Giauque, W.F., and D.P. MacDougall. 1933. Attainment of temperatures below 1[degrees] absolute by demagnetization of [Gd.sub.2][([SO.sub.4]).sub.3] 8[H.sub.2]O. Phys. Rev. 43:7768.
Green, G., G. Patton, J. Stevens, and J. Humphrey. 1986. Reciprocating magnetic refrigerator. Proceedings of the Fourth International Cryocoolers Conference, Easton MD, pp. 65-77.
Gschneidner, K.A., and V.K. Pecharsky. 2000. Influence of magnetic field on the thermal properties of solids. Material Science and Eng. A 287(2):301-10.
Gschneidner, K.A., V.K. Pecharsky, and A.O. Tsokol. 2005. Recent developments in magnetocaloric materials. Rep. Prog. Phys. 68:1479-1539.
Guggenheim, E.A. 1967. Thermodynamics, an Advanced Treatment for Chemists and Physicists. North Holland: Amsterdam.
Hirano, N., S. Nagaya, M. Takahashi, T. Kuriyama, K. Ito, and S. Nomura. 2002. Development of magnetic refrigerator for room temperature application. Advances in Cryogenic Engineeering 47:1027-34.
Lu, D.W., X.N. Xu, H.B Wu, and X. Jin. 2005. A permanent magnet magneto-refrigerator study on using Gd/Gd-Si-Ge/Gd-Si-Ge-Ga alloys. First International Conference on Magnetic Refrigeration at Room Temperature, September 27-30, Montreux, Switzerland.
Nikitin, S.A., A.A. Andreyenko, A.M. Tishin, A.M. Arkharov, and A.A. Zherdev. 1985. Magnetocaloric effect in rare-earth alloys Gd-Ho and Gd-Er. Physics of Metals and Metallography 59(2):104-108.
Okamura, T., K. Yamada, N. Hirano, and S. Nagaya. 2006. Performance of a room-temperature rotary magnetic refrigerator. International Journal of Refrigeration 29(8):1327-31.
Pawlik, K., I Skorvanek, J. Kovac, P. Pawlik, J. Wyslocki, and O.I. Bodak. 2006. Phase structure and magnetocaloric effect in binary Pr-Fe alloys. Journal of Magnetism and Magnetic Materials 304:e510-e512.
Pecharsky, A.O., Gschneidner, K.A. and V.K. Pecharsky. 2003. The giant magnetocaloric effect of optimally prepared [Gd.sub.5][Si.sub.2][Ge.sub.2]. Journal of Applied Physics 93(8):4722-28.
Pecharsky, V.K., and K.A. Gschneidner. 1997a. Giant magnetocaloric effect in [Gd.sub.5]([Si.sub.2][Ge.sub.2]). Physical Review Letters 78(23):4494-97.
Pecharsky, V.K., and K.A. Gschneidner. 1997b. Tunable magnetic regenerator alloys with a giant magnetocaloric effect for magnetic refrigeration from ~20 to ~290 K. App. Phys. Lett. 70(24):3299-3301.
Pecharsky, V.K., and K.A. Gschneidner, Jr. 1997c. Effect of alloying on the giant magnetocaloric effect in [Gd.sub.5]([Si.sub.2][Ge.sub.2]). Journal of Magnetism and Magnetic Materials 167:L179-L184.
Rice, K. 2006. DOE/ORNL heat pump design model, Mark VI Version. www.ornl.gov/~wlj/hpdm/ MarkVI.shtml.
Rowe, A., and A. Tura. 2006. Experimental studies of near room-temperature mangetic refrigeration. International Journal of Refrigeration 29(8):1286-93.
Rowe, A., A. Tura, M.A. Richard, R. Chahine, and J. Barclay, 2004, An overview of operating expereince using the AMR test apparatus. Advances in Cryogenic Engineering 49:1721-28.
Russek, S.L., and C.B. Zimm. 2006. Potential for cost-effective magnetocaloric air-conditioning systems. International Journal of Refrigeration 29(8):1366-73.
Shull, R.D., V. Provenzano, A.J. Shapiro, A. Fu, M.W. Lufaso, J. Karapetrova, G. Kletetschka, and V. Mikula. 2006. The effects of small metal additions (Co, Cu, Ga, Mn, Al, Bi, Sn) on the magnetocalric properties of the [Gd.sub.5][Si.sub.2][Ge.sub.2] alloy. Paper no. 08K908. Journal of Applied Physics 99:1-3.
Tegus, O., E. Brueck, X.W. Li, L. Zhang, W. Dagula, F.R. de Boer, and K.H.J. Buschow. 2004. Tuning of the magnetoc-caloric effects in MnFe(P,As) by substitution of elements. Journal of Magnetism and Magnetic Materials 272-76:2389-90.
Tishin, A.M. 2005. Physical mechanisms of large magnetocaloric effect. First International Conference on Magnetic Refrigeration at Room Temperature, September 27-30, Montreux, Switzerland.
Vasile, C., and C. Muller. 2006. Innovative design of a magnetocaloric system. International Journal of Refrigeration 29(8):1318-26.
Wada, H., C. Funaba, T. Asano, M. Ilyn, and A.M. Tishin. 2005. Recent progress of magnetocaloric effect of [MnAs.sub.1-x][Sb.sub.x]. First International Conference on Magnetic Refrigeration at Room Temperature, September
27-30, Montreux, Switzerland.
Wada, H., and Y. Tanabe. 2001. Giant magnetocaloric effect of [MnAs.sub.1-x][Sb.sub.x]. Applied Physics Letters 79(20):3302-3304.
Wu, W. 2003. Room temperature magnetic refrigerator using a 1.4 T permanent magnet field. American Physical Society Conference, March 3-7, Austin, TX, Source Abstract K7.004.
Yan, A., K.H. Mueller, L. Schultz, and O. Gutfleisch. 2006. Magnetic entropy change in melt-spun MnFePGe. Paper no. 08K903. Journal of Applied Physics 99:1-3.
Yu, B.F., Q. Gao, C.F. Wang, B. Zhang, D.X. Yang, and Y. Zhang. 2006. Experimental investigation on refrigeration performance of a reciprocating active magnetic regenerator of room temperature magnetic refrigeration. International Journal of Refrigeration 29(8):1274-85.
Yu, B.F., Q. Gao, B. Zhang, X.Z. Meng, and Z. Chen. 2003. Review on research of room temperature magnetic refrigeration. International Journal of Refrigeration 26:622-36.
Zhang, Z.Y., Y. Long, R.C. Ye, Y.Q. Chang, and W. Wu. 2005. Corrosion resistance of magnetic refrigerant
gadolinium in water. First International Conference on Magnetic Refrigeration at Room Temperature, September 27-30, Montreux, Switzerland.
Zhou, K.W., Y.H. Zhang, J.Q. Li, J.Q. Deng, and Q.M. Zhu. 2006. Magnetocaloric effects in ([Gd.sub.1-x][Tb.sub.x])[Co.sub.2]. Solid State Communications 137:275-77.
Zimm, C., A. Boeder, J. Chell, A. Sternberg, A. Fujita, S. Fujieda, and K. Fukamichi. 2006. Design and
performance of a permanent magnet rotary refrigerator. International Journal of Refrigeration 29(8):1302-1306.
Zimm, C.B., V.K. Pecharsky, K.A. Gschneidner, Jr., S.A. Nikitin, and A.M. Tishin. 2003. Personal communication with Astronautics Inc., Madison, WI, and Ames Laboratory, Ames, IA.
Kurt L. Engelbrecht
Greg F. Nellis, PhD Member ASHRAE
Sanford A. Klein, PhD Fellow ASHRAE
Carl B. Zimm
Kurt L. Engelbrecht is a graduatestudent, Greg F. Nellis is an assistant professor, and Sanford A. Klein is a professor department, University of Wisconsin-Madison, Madison, WI. Carl B. Zimm is a systems of America, Madison, WI.
|Printer friendly Cite/link Email Feedback|
|Author:||Engelbrecht, Kurt L.; Nellis, Greg F.; Klein, Sanford A.; Zimm, Carl B.|
|Publication:||HVAC & R Research|
|Date:||Jul 1, 2007|
|Previous Article:||The energy crisis--the need for more balanced solutions.|
|Next Article:||Refrigerant distribution in minichannel evaporator manifolds.|