JID:PLA AID:23482 /SCO Doctopic: Condensed matter

[m5G; v1.161; Prn:22/10/2015; 9:34] P.1 (1-6)

Physics Letters A ()

67

Contents lists available at ScienceDirect

68

69

Physics Letters A

70

71

72

www.elsevier.com/locate/pla

73

74

75

10

76

11
12
13

77

An indirect approach based on ClausiusClapeyron equation to

determine entropy change for the rst-order magnetocaloric materials

78
79

14
15
16
17
18
19

80

Kun Xu

a,

, Zhe Li , Yuan-Lei Zhang , Chao Jing

81
82

83

Department of Physics and Electronic Engineering, Key Laboratory for Advanced Functional and Low Dimensional Materials of Yunnan Higher Education
Institute, Qujing Normal University, Qujing 655011, China
b
Department of Physics, Shanghai University, Shanghai 200444, China

84
85

20
21
22
23
24
25
26
27
28
29
30
31
32
33

86

a r t i c l e

i n f o

Article history:
Received 14 September 2015
Received in revised form 15 October 2015
Accepted 15 October 2015
Available online xxxx
Communicated by L. Ghivelder
Keywords:
Heusler alloy
Magnetocaloric effect
ClausiusClapeyron equation
Maxwell relation

87

a b s t r a c t

88

Taking into account the phase fraction during the structural transition for the rst-order magnetocaloric
materials, an improved isothermal entropy change ( S T ) determination has been put forward based
on the ClausiusClapeyron (CC) equation. It was found that the  S T value evaluated by this method
is in excellent agreement with those determined from the Maxwell relation (MR) using magnetic
measurements for some Heusler alloys with a weak eld-induced phase transforming behavior, such
as NiMnSn Heusler alloys. In comparison with the MR based on isoeld magnetization measurements
(MRIF), this method is very convenient to obtain the  S T derived from only few thermomagnetic curves.
More importantly, it is quite superior to the MR-based method in eliminating the overestimation of
 S T due to the appearance of the spurious spike derived from MR employing isothermal magnetization
measurements (MRIT).
2015 Published by Elsevier B.V.

89
90
91
92
93
94
95
96
97
98
99

34

100

35

101

36

102

37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59

1. Introduction
The magnetocaloric effect (MCE), which reects the magnetoresponsive ability of refrigerating when a magnetic material is submitted to the external magnetic eld variations, has emerged as an
attractive option in the new generation of energy-ecient cooling
technologies. Since the discovery of pseudo-binary Gd5 Si2 Ge2 compound by Pecharsky and Gschneidner in 1997 [1], the MCE became
a promising contender to the conventional vapor-compressive refrigerating methods under ambient conditions due to its environmental friendliness, higher cooling eciency and compactness. Since then, the giant MCE has also been extensively studied in other intermetallic compounds, such as MnAs1x Sbx [2],
MnFeP1x Asx [3], (Mn,Fe)2 P1x Six [4], La(Fe1x ,Six )13 /its hydrides
[5,6], and the family of Heusler alloys like NiMnGa [7] and Ni
MnZ (Z = In, Sn, Sb) [8]. Among these materials, the outstanding MCE is due to the involvement of latent heat produced by
structural transition in combination with magnetic ordering and
changes in the electronic band structure. In general, for the materials with direct MCE, both the lattice entropy change ( S L ) and
magnetic entropy change ( S mag ) contribute positively to the total measured MCE [9]. Whereas for the materials with inverse MCE,
the magnetocaloric response comes from the  S L , and the  S mag

60
61
62
63
64
65
66

Corresponding author.
E-mail address: xukun0830@hotmail.com (K. Xu).

http://dx.doi.org/10.1016/j.physleta.2015.10.029
0375-9601/ 2015 Published by Elsevier B.V.

plays a negative role [10,11]. Thus, entropy change associated with

the rst-order magnetostructural phase transition (FOMT) tends to
be much more complicated than the second-order, i.e., purely magnetic transition.
As is well known, the characteristic parameters for a magnetocaloric compound are the isothermal entropy change ( S T ) and
the adiabatic temperature change ( T ad ) when a magnetic eld is
applied or removed. To fully exploit the MCE for a given material, it
is essential to clarify the evolution of  S T or  T ad with temperature and magnetic eld. Usually, determining the  T ad with direct
or quasi-direct calorimetric methods is dicult and challenging.
Therefore, the simplest and most widely used experimental approach has been put forward, i.e., the  S T can be derived from
Maxwell relation with a numerical form using isothermal magnetization measurements (denoted as MRIT here)

H 2
ST =
H1

 M i +1 M i
M
H .
dH
T
T i +1 T i

103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119

(1)

However, the MRIT has encountered some challenges during

the last two decades. For example, a large experimental discrepancy led Gigure et al. [12] to claim that, rather than the MRIT,
the ClausiusClapeyron (CC) equation should apply to calculate
the MCE associated with FOMT, since its magnetization curve is
not a continuous, derivable function. Soon after, Gschneidner et
al. [13] argued that the discrepancy just originates from kinetic

120
121
122
123
124
125
126
127
128
129
130
131
132

JID:PLA

AID:23482 /SCO Doctopic: Condensed matter

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41

[m5G; v1.161; Prn:22/10/2015; 9:34] P.2 (1-6)

K. Xu et al. / Physics Letters A ()

effect. Meanwhile, Sun et al. [14] also pointed out that CC equation is just a special case of integrated Maxwell relation and is
inadequate to evaluate the  S T value for an incomplete transformation. Particularly, recent progress on the inverse MCE research for NiMnSn alloys aroused a new round of argument once
again. Zou et al. [15,16] claimed that the  S T value is seriously
overestimated by MRIT in Ni43 Mn46 Sn11 due to the occurrence
of a magnetically inhomogeneous martensitic state. On the contrary, Maosa et al. [17] proved that the  S T values respectively
obtained from MRIT and calorimetric measurements are in well
agreement for Ni50 Mn35 Sn15 . Despite the validity and generality of
MRIT being still under debate, it is undeniable that the direct application of MRIT would lead to some unphysical results. In most
cases, a spike is always observed on the  S T ( T ) curve, especially in some materials with a pronounced eld-induced transition
[2,1822]. A spike value for Mn0.997 Fe0.003 As unexpectedly attains
up to 320 J/kg K for a eld change of 5 T under the ambient pressure [23]. This value is far above those reported for all
known magnetocaloric materials as well as the theoretical prediction. Such a colossal MCE can be attributed to the incorrect
application of MRIT, which gives rise to a spurious artifact. In fact,
some practical methods to modify the  S T values obtained from
MRIT have been proposed by Liu et al. [24] and Cui et al. [25].
Nevertheless, the former one was found to be inadequate for an
extreme case of Mn0.99 Cu0.01 As [26], while the latter one is very
complex for the practical use. In addition to MRIT, the  S T can
be also determined accurately by using Maxwell relation based on
isoeld magnetization measurements (referred to as MRIF here)
[27], but this method needs a large number of thermomagnetic
curves with great measuring cost. In the present work, a proper
method based on the CC equation using isoeld magnetization
measurements (CCIF) has been proposed to estimate the realistic
MCE for the magnetocaloric materials with FOMT. The feasibility
of this method was tested on the Ni43 Mn46 Sn11 , Ni50 Mn35 Sn15 and
Ni45 Co5 Mn36.6 In13.4 Heusler alloys with their inverse MCE, as well
as MnAs and Gd5 Si2 Ge2 compounds with direct MCE. All these experimental results demonstrate that such a new method using to
evaluate the  S T is superior to the MR-based method.

The isothermal magnetization measurements were performed

as: (i) the isothermal M ( H ) curve measured on heating mode:
heating the sample at zero elds from T to the measuring temperature T i without overshooting the target temperature, and then
recording the isothermal magnetization M ( H ) by scanning elds
(0 H , T i ). Next, the sample was cooled back down to T at zero
elds, and heated to the next temperature, T i +1 , for measuring.
(ii) Isothermal M ( H ) curve measured on cooling mode: cooling
the sample down from T + to T i without undershooting the target temperature, and collecting the isotherm M ( H ) by scanning
elds (0 H , T i ). Afterwards, the sample was heated up to T +
at zero elds and cooled down to the next temperature, T i +1 , for
measuring. Before each M ( H ) curve measurement, the sample was
carefully recovered to the initial state, since the rst-order system
could remain partially transformed in higher magnetization after
removing the saturate eld if the cooling martensitic transformation is not crossed [26,29].

2. Experimental

3. Results and discussion

42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66

Nominal stoichiometry Ni43 Mn46 Sn11 and Ni50 Mn35 Sn15 ingots
were prepared by arc melting high-purity Ni, Mn, Sn metals under
an argon atmosphere. After melting, the samples were sealed in
evacuated quartz tubes in order to homogenize at 1173 K for 24 h,
and then subsequently quenched in ice-water. Their crystal structures and phases were veried by the X-ray diffraction (Rigaku,
Ultima IV) using Cu K radiation. The results show that both the
Ni43 Mn46 Sn11 and Ni50 Mn35 Sn15 samples are single phase with a
L21 cubic structure (austenite phase) at room temperature. The
magnetic characterizations were carried out by a Quantum Design VersaLab magnetometer (VersaLab, 3 Tesla). When dealing
with the Ni45 Co5 Mn36.6 In13.4 , MnAs and Gd5 Si2 Ge2 samples, the
correlatively experimental data were directly extracted from previous literatures [2,22,28].
The isoeld magnetization measurements were carried out as:
cooling the sample down to T (a full martensitic state) in the
absence of magnetic elds, and then the measurement was made
on increasing temperature from the T to T + (a full austenitic
state) with a scan rate of 1.0 K/min in the presence of magnetic
eld (H , T T + ). In this way, the isoeld M ( T ) heating curve
was obtained. Subsequently, the measurement was taken on the
decreasing temperature from the T + to the T without removing
the magnetic eld (H , T + T ), and the isoeld M ( T ) cooling
curve was acquired as well.

67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82

Fig. 1. The M ( T ) curve under a static magnetic eld of 1 T for Ni43 Mn46 Sn11 sample
(full circles). The red and blue dot lines represent M A ( T ) and M M ( T ) obtained by
extrapolating the linear part of magnetization. The corresponding austenite fraction
f ( T ) is represented using a green solid line. (For interpretation of the references to
color in this gure legend, the reader is referred to the web version of this article.)

83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107

In order to illustrate the use of the CCIF method, we rstly

selected the intensively studied ternary Ni43 Mn46 Sn11 Heusler alloy as a representative. Fig. 1 shows its M ( T ) heating curve from
T = 150 K to T + = 240 K in the presence of a static eld H = 1 T.
The sample undergoes a reverse martensitic transformation from a
weak-magnetic martensite to a ferromagnetic austenite phase. To
our knowledge, the nite transition width is due to the inhomogeneous nature of the sample [10], and the transforming process can
be supposed to be continuous. In this case, austenite phase fraction, f ( T ), should be taken into account and it can be described
by using M ( T ) data. This is based on an assumption that the total magnetization is proportional to the phase volume fraction [10,
3033]. It is worth noting that the kinetic arrest effect [34] which
only occurs at very high elds is ignored here. In order to determine the f ( T ), it is necessary to rst obtain the magnetization
of full martensite M M ( T ) and austenite phases M A ( T ), which can
be roughly deduced by linearly extrapolating heating curve outside
the transforming region, as illustrated by the blue/red dot lines in
Fig. 1. By combining M M ( T ) and M A ( T ), the f ( T ) can be calculated by the below expression

f (T ) =

M (T ) M M (T )
M A (T ) M M (T )

108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128

(2)

Following Eq. (2), the resultant f ( T ) curve at H = 1 T was obtained, as indicated by the green solid line in Fig. 1.

129
130
131
132

JID:PLA AID:23482 /SCO Doctopic: Condensed matter

[m5G; v1.161; Prn:22/10/2015; 9:34] P.3 (1-6)

K. Xu et al. / Physics Letters A ()

67

68

69

70

71

72

73

74

75

10

76

11

77

12

78

13

79

14

80

15

81

16
17
18
19
20

82

Fig. 2. The schematic illustration of calculated f ( T ) (the black dashed line for H =
50 mT, and the green one for H = 3 T) and the corresponding  f ( T ) (red solid
line) curves in the heating run for Ni43 Mn46 Sn11 sample. Inset: The M ( T ) heating
curve under elds of 50 mT and 3 T. (For interpretation of the references to color
in this gure legend, the reader is referred to the web version of this article.)

83
84
85
86

21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49

87

In consideration of the fact that both the temperature and the

magnetic eld are equivalent thermodynamically as a driving force
for the phase transition [33], we believe that the impact of magnetic eld on the f ( T ) can be quantitatively estimated by using
M ( T ) curves measured under different magnetic elds. The inset
of Fig. 2 shows the M ( T ) curves with heating mode under 50 mT
and 3 T, respectively. From this gure, it is found that only a partial
phase fraction can be arrested by applying a moderate eld owing
to a broad transforming region (25 K) for the studied sample. In
order to obtain the exact amount of the change in f ( T ) triggered
by a change of magnetic eld ( H ) from an initial eld (H i =
50 mT) to a nal eld (H f = 3 T), we adopted the same approach
and plotted the corresponding f ( T ) curves at these two magnetic
elds as shown in Fig. 2. Consequently, the change of phase fraction induced by the change of magnetic eld,  f ( T ,  H ), can be
computed using the expression  f ( T ,  H ) = f ( T , H f ) f ( T , H i ),
as plotted by the red line in Fig. 2. Moreover, the total entropy
changes ( S tr ) across full transition can be quantitatively estimated by CC equation [35,36]

 S tr =  M (dT t /dH )1 .

52
53
54
55
56
57
58
59
60
61
62
63
64
65
66

89
90
91

Fig. 3. (a) The M ( T ) isoeld curves upon heating and cooling in magnetic elds
of 50 mT, 1 T, 2 T and 3 T. (b) T t versus H phase diagram deduced from isoeld
curves. (c) Field dependence of  M for Ni43 Mn46 Sn11 .

 S T  f  S tr =  f  M

dT t

93
94
96
97
98
99
100
101
102
103
104
105
106
107

(3)

dH

92

95

108
109

Here, the  M is the difference of magnetization between the

two phases, while the dT t /dH is the change rate of phase transition equilibrium temperature with magnetic eld. Because the
 f ( T ,  H ) directly reects the phase transformation completion
percent induced by a change of magnetic eld, evaluation of the
 S T with the CCIF method has been well established as the following equation,

50
51

88

110
111
112
113
114
115
116

117

(4)

For Ni43 Mn46 Sn11 alloy, the M ( T ) curves at different magnetic

elds are presented in Fig. 3(a). It is clearly observed that the T t
(the middle point in magnetization change) almost linearly decreases with the increase of magnetic elds. The values of dT t /dH
are about 2.03 K/T upon heating and 2.28 K/T upon cooling
(see Fig. 3(b)). From Fig. 3(c), one can notice that the  M almost
keeps constant with increasing magnetic elds. The obtained average values are 50.7 emu/g on heating and 51.0 emu/g on
cooling. Substituting these parameters into Eq. (3), the values of
 S tr for Ni43 Mn46 Sn11 are calculated to be 24.97 J/kg K and
22.37 J/kg K under heating/cooling modes, respectively. Moreover,
the  f ( T ,  H ) from H i (50 mT) to H f (1 T, 2 T, 3 T) can be
computed by subtracting two corresponding f ( T ) curves. According to Eq. (4), the  S T ( T ) curves are plotted in Figs. 4(a) and (b).

118
119
120
121
122
123

Fig. 4. The calculated  S T values for Ni43 Mn46 Sn11 : (a)(b) by CCIF; (c)(d) by
MRIT; (e)(f) by MRIF. The left panel shows those upon cooling, and the right panel
shows those for heating.

124
125
126
127

To check the validity of CCIF method, the  S T ( T ) curves are also

calculated by using magnetic measurements based on Maxwell relation (i.e. MRIT and MRIF) for comparison, as shown in Figs. 4(c),
(d), (e) and (f). Obviously, the spurious spike does not appear in
these results. This behavior can be ascribed to an insucient Zee-

128
129
130
131
132

JID:PLA

AID:23482 /SCO Doctopic: Condensed matter

[m5G; v1.161; Prn:22/10/2015; 9:34] P.4 (1-6)

K. Xu et al. / Physics Letters A ()

67

68

69

70

71

72

73

74

75

10

76

11

77

12

78

13

79

14

80

15

81

16

82

17

83

18

84

19

85

20

86

21

87

22

88

23

89

24

90

25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66

91

Fig. 5. (a) The M ( T ) isoeld curves upon heating and cooling in magnetic elds of
50 mT, 0.5 T, 1 T, 2 T and 3 T. (b) T t versus H phase diagram deduced from isoeld
curves. (c) Field dependence of  M for Ni50 Mn35 Sn15 . (For interpretation of the
references to color in this gure, the reader is referred to the web version of this
article.)

man energy supplied by the magnetic eld, which cannot promote

a complete transformation, resulting in an unsaturated  S T for
Ni43 Mn46 Sn11 . Therefore, each  S T ( T ) curve deduced from these
methods shows a peak shape, and their peak value are well coincident (13 J/kg K for a eld change of 3 T). Such a comparison
highlights the fact that these methods in estimating  S T values
are quite effective. However, for the cooling protocol, the peak positions derived from CCIF and MRIF are relatively lower (3 K)
than that obtained from MRIT. This small discrepancy probably
originates from the intrinsic hysteresis effect for the studied sample.
Next, the same approach was implemented on another ternary
Heusler alloy of Ni50 Mn35 Sn15 , where T and T + were chosen
to be 120 K and 230 K, respectively. Compared to Ni43 Mn46 Sn11 ,
a larger thermal hysteresis (20 K) and a smaller saturate  M
(28.0 emu/g upon heating and 31.0 emu/g upon cooling) are
remarkably displayed in Figs. 5(a) and (c). From Fig. 5(b), however,
the linear relation between T t and H drastically deviates in the
eld range below 0.3 T. Such a changing tendency has also been
reported in previous works [31,37]. It undoubtedly manifests that
the CCIF is incapable to evaluate  S T for Ni50 Mn35 Sn15 over the
whole eld span. To solve this problem, we selected both the H i
and the H f in the linear region (see blue solid line) and collected
the dT t /dH with the values of 1.47 K/T and 1.75 K/T
upon heating and cooling, respectively. Utilizing these experimental parameters, the corresponding  S tr calculated by Eq. (3) are
19.05 J/kg K and 17.71 J/kg K, respectively. These values are signicantly smaller than those in Ni43 Mn46 Sn11 sample. On the basis
of CCIF method, the temperature dependence of  S T for the eld
change from H i (0.5 T) to H f (1 T, 2 T and 3 T) was presented in
Figs. 6(a) and (b). It also presents some large peaks around transformation point. In contrast with Ni43 Mn46 Sn11 alloy, however,
the maximum  S T for the eld change from H i = 0.5 T to H f =
3 T in this sample rapidly decreases to 7 J/kg K and 9 J/kg K
during heating/cooling mainly due to reduction of the  S tr . For

92
93
94

Fig. 6. The calculated  S T values for Ni50 Mn35 Sn15 : (a)(b) by CCIF; (c)(d) by
MRIT; (e)(f) by MRIF. The left panel shows those upon cooling, and the right panel
shows those for heating.

95
96
97
98

Ni50 Mn35 Sn15 alloy, more importantly, the  S T ( T ) curves calculated by every method (CCIF, MRIF and MRIT) are all coincident
(see Fig. 6), and are also in well agreement with that obtained
by quasi-direct calorimetric measurements [17]. These experimental results further prove that the CCIF method proposed by us to
determine MCE in NiMnSn alloys is faithworthy. According to
previous works [10,35], we can nd that an inverse MCE in Mnrich NiMn based Heusler alloys should primarily rely on three parameters including the  M, the dT t /dH and the transition width.
Considering the fact that the magnetic contribution works against
the structural contribution in the NiMnSn system, it is necessary
to decrease the  M suitably, and simultaneously maintain a strong
coupling to the magnetic eld for obtaining an optimized MCE.
Under this condition, in terms of Eq. (3), a larger  S tr requires
a smaller dT t /dH . In this case, the only a partial phase transition
(corresponding to a small  f ) would be arrested by magnetic eld
in the presence of wide transition width. For example, the dT t /dH
with a value of 2 K/T in Ni43 Mn46 Sn11 alloy, applying a magnetic
eld change of 3 T can only shift its f ( T ) curve by 6 K, which is
signicantly lower than that of transition width (25 K). As a consequence, only 50% of the transition can be induced by the eld
(see in Fig. 2) and resultant  S T will thus be decreased dramatically in comparison to its  S tr (see in Fig. 4). Therefore, we can
conclude that a relatively low dT t /dH is accompanied with a broad
transition width for these two NiMnSn samples, which leads to
the fact that  f can only maintain in a narrow temperature interval, thus giving rise to a peak-shape  S T ( T ) curves (see in Figs. 4
and 6). In reality, such a phenomenon actually reects the intrinsic nature of inverse MCE for NiMnSn system with martensitic
transformation.
A great deal of experimental studies indicated that the introduction of suitable transition-metal (Co, Fe, etc.) can improve
the inverse MCE of ternary NiMnZ alloys [22,3840]. For instance, Chen et al. have obtained an enhanced MCE in quaternary

99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132

JID:PLA AID:23482 /SCO Doctopic: Condensed matter

[m5G; v1.161; Prn:22/10/2015; 9:34] P.5 (1-6)

K. Xu et al. / Physics Letters A ()

67

68

69

70

71

72

73

74

75

10

76

11

77

12

78

13

79

14

80

15

81

16
17
18
19
20
21

82

Fig. 7. The  S T curves for a eld change  H = 5 T of Ni45 Co5 Mn36.6 In13.4 by CCIF
(red open circles) and MRIT (black solid rectangles). The spike (shadow area) corresponding to MRIT was removed by the authors of Ref. [22]. (For interpretation
of the references to color in this gure legend, the reader is referred to the web
version of this article.)

Fig. 8. The  S T curves of MnAs by both CCIF (solid lines) and MRIT (dot lines). The
original M ( T ) data and  S T results derived from MRIT are extracted from Ref. [2].

24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50

86
87
88

Ni45 Co5 Mn36.6 In13.4 alloy. Owing to the optimization of correlative

parameters ( M = 97.49 emu/g and dT t /dH = 5.0 K/T) for the
studied sample, an extremely high spike has been observed in its
 S T ( T ) curve on account of the prominent eld-induced martensitic transition, as revealed by the square line in Fig. 7. Notably,
the spike value attains to 33.4 J/kg K for a eld change from 0
to 5 T, which exceeds the value of  S tr (19.5 J/kg K). Such a result implies that the spike in  S T ( T ) curve is an unphysical result
due to the overestimation by using MRIT. In fact, the authors also
deemed that an effective  S T of this sample should be located
at the inection point, and the spurious spike was removed articially from the  S T ( T ) curve (see in the shadow area of Fig. 7).
By applying CCIF, the  S T ( T ) for the same magnetic eld change
is reconsidered in Ni45 Co5 Mn36.6 In13.4 , as shown in cycle line of
Fig. 7. More interestingly, it is found that the nal result from CCIF
is in nicely agreement with the spurious spike-removed curves
calculated from MRIT, and the problem of overestimation in the
 S T ( T ) curve is eliminated effectively.
In addition to Heusler alloys, we also tried to generalize the
CCIF for the typical giant magnetocaloric materials, which usually
show a direct MCE in the vicinity of FOMT. Herein, the MnAs,
as a representative compound, was selected to test the generality of this method. For the MnAs compound, as being distinct from
aforementioned Heusler alloys, it often experiences a very sharp
transformation (with transition width less than 5 K) from a paramagnetic state with the orthorhombic NiAs-type structure to a
ferromagnetic state with the hexagonal MnP-type structure in the
cooling process [2]. In this case, Eq. (2) must be revised as follows:

51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66

84
85

22
23

83

f (T ) =

M ( T ) M LT ( T )
M HT ( T ) M LT ( T )

(5)

Here, the M LT ( T ) and M HT ( T ) denote the magnetizations of lowand high-temperature phases, respectively. Adopting the  M and
the dT t /dH with the values of 95.5 emu/g and 3.4 K/T, the  S T of
MnAs in various magnetic elds were recalculated by using CCIF,
as displayed in Fig. 8. From this gure, it can be remarkably distinguished that the maximum  S T for this sample is 28.09 J/kg K,
rather than the spike value (40.8 J/kg K) calculated by the authors. The appearance of spurious spike has also been observed
in other giant magnetocaloric compounds [28,41], which means
that the MRIT is incapable used in the case of mixed phase near
transition point. Consequently, from the results calculated by us,
the  S T quickly reaches a saturated value at a low magnetic eld,
while the further increasing magnetic eld only plays a pivotal

89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114

Fig. 9. (a) The  S T curves of Gd5 Si2 Ge2 by MRIT, CCIF and another M ( T )-based
method in Ref. [28]. (b) The background removed  S T curves from these three
methods. (c) The  S T curve by CCIF with consideration of the variation of  M.
Insets: The eld dependences of T t and  M.

115
116
117
118
119

role for broadening of working temperature range, reecting the

intrinsic nature of MCE in MnAs with its FOMT. These ndings also
pronounce that the CCIF applying for a giant magnetocaloric system is very credible.
In the end of this paper, we adopted this method on another
important system with FOMT, Gd5 Si2 Ge2 . The same as MnAs, a direct MCE in Gd5 Si2 Ge2 is also associated with a prominent eldinduced phase transition, resulting in a spurious spike presented
in its  S T ( T ) curve by the MRIT method [28]. Here, we extracted
the necessary data in Gd5 Si2 Ge2 from Ref. [28]. From the insets of
Fig. 9, in the case of studied compound, it can be clearly found
that its  M reaches saturation (85.9 emu/g) rapidly followed
by a monotonous decrease with further increasing eld strength,

120
121
122
123
124
125
126
127
128
129
130
131
132

JID:PLA

AID:23482 /SCO Doctopic: Condensed matter

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29

while its T t versus H curve presents a linear relationship above

0.4 T with a slope of 6.064 K/T. Using these two parameters,
we calculated the  S T (0.4 T 5 T) according to CCIF, which
is compared with the results estimated by MRIT (0.4 T 5 T) and
another M ( T )-based method (0 T 5 T) in Ref. [28], as shown in
Fig. 9(a). Unexpectedly, the  S T ( T ) curve obtained by CCIF deviates from MRIT, and the areas under these two  S T curves are also
different. It contradicts the fact that the area under  S T curves
at the same condition should be the same [26,28]. By a carefully
observation on the  S T values outside the transition range, the
 S T from MRIT maintains a nonzero value, while the  S T from
CCIF has been reset to zero in the process of normalization by
Eq. (5). The former behavior originates from the weak magnetization variation with temperature, which leads to the discrepancies
in the areas under these two  S T curves. As shown in Fig. 9(b),
the background of  S T curve deduced by MRIT was removed, and
the area under this curve reduces greatly from 537.4 J/kg to a value
of 354.3 J/kg. This value becomes smaller than that calculated by
CCIF (390.2 J/kg) in turn, indicating that the integrated  S T from
CCIF seems to be overestimated. For the  S T evaluated by CCIF in
Figs. 9(a) and (b), the parameter  M has been chosen to be the
maximum value. Considering the fact that both the  M and the
 S tr actually decrease with the magnetic eld, the method of segmentation has to be adopted to re-evaluate the  S T by CCIF. The
magnetic eld region can be divided into several sections. For each
section, both the  M and the dT t /dH are approximately viewed as
constant, and the related  S Ti can be determined by CCIF directly.
Consequently, the total  S T can be easily calculated through the
accumulation of  S Ti using the following equation:

30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53

[m5G; v1.161; Prn:22/10/2015; 9:34] P.6 (1-6)

K. Xu et al. / Physics Letters A ()

ST

n


 S Ti

n

i

 f i M i

dH
dT t

(6)

56
57
58
59
60
61
62
63

[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]

[25]
[26]

4. Conclusions

[28]

In summary, the method based on CC-equation while accounting for the phase fraction change was developed to determine
the  S T for some representative materials, such as NiMnSn,
NiCoMnIn, MnAs and Gd5 Si2 Ge2 . Our experimental results unambiguously demonstrated that this method is a quite simple and
effective way to obtain a realistic value of  S T in the particularly interesting class of magnetocaloric materials. In comparison
to MRIF, moreover, this method only requires a few M ( T ) curves
and reduces the cost of measuring signicantly. Such a method
should be helpful to equitably deal with the MCE for many magnetic materials during FOMT.

[29]

[24]

[27]

[30]
[31]
[32]
[33]
[34]
[35]
[36]

Acknowledgements
[37]

This work was supported by the National Natural Science Foundation of China (Grant Nos. 11364035, 11404186, and 51371111),
the Key Basic Research Program of Science and Technology Commission of Shanghai Municipality (Grant No. 13JC1402400), Project
for Innovative Research Team of Qujing Normal University (Grant
No. TD201301), and Project for Applied Basic Research Programs of
Yunnan Province (Grant Nos. 2013FZ110 and 2012FD051).

67
68

[1]
[2]
[3]
[4]

Accordingly, the eld range from 0.4 T to 5 T was roughly divided

into ve sections, and their  S Ti curves were shown in Fig. 9(c). By
using the above equation, the area under CCIF curve is reduced to
355.9 J/kg, which is coincident excellently with MRIT (354.3 J/kg).
Therefore, such a CCIF method based on Eq. (6) could be extended
to more general cases, even if the  M and dT t /dH are not constant.

54
55

References

[38]
[39]
[40]
[41]

V.K. Pecharsky, K.A. Gschneidner Jr., Phys. Rev. Lett. 78 (1997) 4494.
H. Wada, Y. Tanabe, Appl. Phys. Lett. 79 (2001) 3302.
O. Tegus, E. Brck, K.H.J. Buschow, F.R. de Boer, Nature 415 (2002) 150.
X.F. Miao, L. Caron, P. Roy, N.H. Dung, L. Zhang, W.A. Kockelmann, R.A. de Groot,
N.H. van Dijk, E. Brck, Phys. Rev. B 89 (2014) 174429.
F.X. Hu, B.G. Shen, J.R. Sun, Z.H. Cheng, G.H. Rao, X.X. Zhang, Appl. Phys. Lett.
78 (2001) 3675.
A. Fujita, S. Fujieda, Y. Hasegawa, K. Fukamichi, Phys. Rev. B 67 (2003) 104416.
M. Pasquale, C.P. Sasso, L.H. Lewis, L. Giudici, T. Lograsso, D. Schlagel, Phys. Rev.
B 72 (2005) 094435.
T. Krenke, E. Duman, M. Acet, E.F. Wassermann, X. Moya, Ll. Maosa, A. Planes,
Nat. Mater. 4 (2005) 450.
V.K. Pecharsky, K.A. Gschneidner Jr., Ya. Mudryk, D. Paudyal, J. Magn. Magn.
Mater. 321 (2009) 3541.
J. Liu, T. Gottschall, K.P. Skokov, J.D. Moore, O. Guteisch, Nat. Mater. 11 (2012)
620.
Z. Li, Y.L. Zhang, K. Xu, T.X. Yang, C. Jing, H.L. Zhang, Solid State Commun. 203
(2015) 81.
A. Gigure, M. Foldeaki, B. Ravi Gopal, R. Chahine, T.K. Bose, A. Frydman, J.A.
Barclay, Phys. Rev. Lett. 83 (1999) 2262.
K.A. Gschneidner Jr., V.K. Pecharsky, E. Brck, H.G.M. Duijn, E.M. Levin, Phys.
Rev. Lett. 85 (2000) 4190.
J.R. Sun, F.X. Hu, B.G. Shen, Phys. Rev. Lett. 85 (2000) 4191.
J.D. Zou, B.G. Shen, B. Gao, J. Shen, J.R. Sun, Adv. Mater. 21 (2009) 693.
J.D. Zou, B.G. Shen, B. Gao, J. Shen, J.R. Sun, Adv. Mater. 21 (2009) 3727.
Ll. Maosa, A. Planes, X. Moya, Adv. Mater. 21 (2009) 3725.
L. Tocado, E. Palacios, R. Burriel, J. Appl. Phys. 105 (2009) 093918.
S. Gama, A.A. Coelho, A. de Campos, A.M.G. Carvalho, F.C.G. Gandra, Phys. Rev.
Lett. 93 (2004) 237202.
Z.G. Guo, L.Q. Pan, M.Y. Raque, X.F. Zheng, H.M. Qiu, Z.H. Liu, J. Alloys Compd.
577 (2013) 174.
T. Krenke, E. Duman, M. Acet, E.F. Wassermann, X. Moya, Ll. Maosa, A. Planes,
E. Suard, B. Ouladdiaf, Phys. Rev. B 75 (2007) 104414.
L. Chen, F.X. Hu, J. Wang, L.F. Bao, J.R. Sun, B.G. Shen, J.H. Yin, L.Q. Pan, Appl.
Phys. Lett. 101 (2012) 012401.
A. de Campos, D.L. Rocco, A.M.G. Carvalho, L. Caron, A.A. Coelho, S. Gama,
L.M. da Silva, F.C.G. Gandra, A.O.D. Santos, L.P. Cardoso, P.J. von Rank, N.A. de
Oliveira, Nat. Mater. 5 (2006) 802.
G.J. Liu, J.R. Sun, J. Shen, B. Gao, H.W. Zhang, F.X. Hu, B.G. Shen, Appl. Phys.
Lett. 90 (2007) 032507.
W.B. Cui, W. Liu, Z.D. Zhang, Appl. Phys. Lett. 96 (2010) 222509.
L. Caron, Z.Q. Ou, T.T. Nguyen, D.T. Cam Thanh, O. Tegus, E. Brck, J. Magn.
Magn. Mater. 321 (2009) 3559.
V.V. Khovaylo, S.P. Skokov, O. Guteisch, H. Miki, T. Takagi, T. Kanomata, V.V.
Koledov, V.G. Shavrov, G. Wang, E. Palacios, J. Bartolom, R. Brurriel, Phys. Rev.
B 81 (2010) 214406.
A.M.G. Carvalho, A.A. Coelho, P.J. van Ranke, C.S. Alves, J. Alloys Compd. 509
(2011) 3452.
G. Porcari, S. Fabbrici, C. Pernechele, F. Albertini, M. Buzzi, A. Paoluzi, J. Kamarad, Z. Arnold, M. Solzi, Phys. Rev. B 85 (2012) 024414.
P.J. Shamberger, F.S. Ohuchi, Phys. Rev. B 79 (2009) 144407.
E. Stern-Taulats, P.O. Castillo-Villa, Ll. Maosa, C. Frontera, S. Pramanick, S. Majumdar, A. Planes, J. Appl. Phys. 115 (2014) 173907.
V. Basso, C.P. Sasso, K.P. Skokov, O. Guteisch, V.V. Khovaylo, Phys. Rev. B 85
(2012) 014430.
B. Emre, S. Yce, E. Stern-Taulats, A. Planes, S. Fabbrici, F. Albertini, Ll. Maosa,
J. Appl. Phys. 113 (2013) 213905.
W. Ito, K. Ito, R.Y. Umetsu, R. Kainuma, K. Koyama, K. Watanabe, A. Fujita, K.
Oikawa, K. Ishida, T. Kanomata, Appl. Phys. Lett. 92 (2008) 021908.
V. Recarte, J.I. Prez-Landazbal, S. Kustov, E. Cesari, J. Appl. Phys. 107 (2010)
053501.
V.K. Pecharsky, K.A. Gshneidner Jr., A.O. Pecharsky, A.M. Tishin, Phys. Rev. B 64
(2001) 144406.
J.M. Barandiaran, V.A. Chernenko, E. Cesari, D. Salas, P. Lazpita, J. Gutierrez, I.
Orue, Appl. Phys. Lett. 102 (2013) 071904.
L. Feng, L. Ma, E.K. Liu, G.H. Wu, W.H. Wang, W.X. Zhang, Appl. Phys. Lett. 100
(2012) 152401.
M. Khan, I. Dubenko, S. Stadler, J. Jung, S.S. Stoyko, A. Mar, A. Quetz, T. Samanta,
N. Ali, K.H. Chow, Appl. Phys. Lett. 102 (2013) 112402.
X.G. Zhao, M. Tong, C.W. Shih, B. Li, W.C. Chang, W. Liu, Z.D. Zhang, J. Appl.
Phys. 113 (2013) 17A913.
B.G. Shen, J.R. Sun, F.X. Hu, H.W. Zhang, Z.H. Cheng, Adv. Mater. 21 (2009) 1.

69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129

64

130

65

131

66

132