Influence of Ni on the magnetocaloric effect in Nanoperm-type soft-magnetic amorphous alloys

Journal of Alloys and Compounds 591 (2014) 29–33

Contents lists available at ScienceDirect

Journal of Alloys and Compounds journal homepage: www.elsevier.com/locate/jalcom

Influence of Ni on the magnetocaloric effect in Nanoperm-type soft-magnetic amorphous alloys B. Podmiljsak a,c,⇑, J.-H. Kim b, P.J. McGuiness a,c, S. Kobe a a

Department for Nanostructured Materials, Jozef Stefan Institute, Jamova cesta 39, 1000 Ljubljana, Slovenia Analysis Research Devision, Korea Basic Science Institute, 1370, Sankyuk-Donk, Buk-gu, Daegu, Republic of Korea c Center of Excellence NAMASTE, Ljubljana, Jamova cesta 39, 1000 Ljubljana, Slovenia b

a r t i c l e

i n f o

Article history: Received 11 July 2013 Received in revised form 13 December 2013 Accepted 16 December 2013 Available online 28 December 2013 Keywords: Amorphous materials Rapid-solidification Magnetization Magnetocaloric effect Refrigerant capacity Curie temperature

a b s t r a c t We have studied the influence of Ni on the magnetocaloric effect (MCE) in Nanoperm-type amorphous materials by investigating a series of Fe84xNixZr6B10 alloys with x = 0, 2, 4 and 6. As expected, the Curie temperature increased with the amount of Ni from 427 K for x = 0 to 482 K for x = 6. The maximum magnetic entropy change ðDSpk M Þ for an applied field of 1.4 T also increased, reaching a value of 1.52 J/K kg for x = 6, which is an increase of 25% compared to the Ni-free alloy. The refrigerant capacity first decreased for x = 2 and then increased, reaching a maximum value of 93 J/kg (DH = 1.4 T) for x = 6. For a 5 T field change, the value increased to 407 J/kg, which is higher than the 355 J/kg achieved with Gd5Ge1.9Si2Fe0.1. We confirmed the proposed ‘‘master curve’’ behavior for second-order magnetic transition (SOMT) alloys for the temperature dependence of DSM for different alloy compositions of the same series, which makes it easier and faster to find proper candidates for a magnetic refrigerator. Ó 2013 Elsevier B.V. All rights reserved.

1. Introduction Iron-based amorphous alloys have received a great deal of attention since their discovery in the 1960s [1], because of their excellent soft-magnetic properties and their consequent use in electronic devices such as transformer cores. However, they have only recently drawn attention as a working material for room-temperature magnetic refrigerators, where they have a number of advantages over already known magnetocaloric materials [2–7]. Magnetic refrigerators work on the principle of the magnetocaloric effect [8], which involves the temperature change of a magnetic material upon the application of a magnetic field. When magnetized, the entropy of the spin subsystem is decreased and, under adiabatic conditions, the transfer of energy to the lattice induces heating of the material. Conversely, the adiabatic demagnetization of the material causes its cooling. Until now, materials with a first-order magnetic transition (FOMT), like Gd5(Si,Ge)4 [9], La(Fe,Si)13 [10], FeMnPAs [11], Ni2(Mn,Ga) [12] and others [13], have been considered the best choice for magnetic refrigerators because of their very large ðDSpk M Þ, which occurs as a result of a coupled magneto-structural transition. But these structural changes also present some draw⇑ Corresponding author at: Department for Nanostructured Materials, Jozef Stefan Institute, Jamova cesta 39, 1000 Ljubljana, Slovenia. Tel.: +386 14773818. E-mail address: [email protected] (B. Podmiljsak). 0925-8388/$ - see front matter Ó 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jallcom.2013.12.150

backs. They have very large thermal and magnetic-field hysteresis, which can be reduced by substituting alternative elements, but only at the expense of a lower DSM [14]. Furthermore, because of the structural transition, large volume changes and stresses between the coexisting phases appear, which make the sample very brittle [15]. The operating temperature range also tends to be very narrow and some of the proposed materials contain toxic elements. Materials with a SOMT lack the very large DSpk M , but they do have a very high refrigeration capacity (RC), which is now recognized as the key parameter, because it is a better way by which to compare different MC materials. The other positive characteristics of SOMT materials are a low magnetic hysteresis, a high electrical resistivity, enhanced corrosion resistance, good mechanical properties, and a TC tunable by varying the composition [3,7]. Fe-based amorphous magnetocalorics also have the advantage of generally being very cheap and easy to produce. What tends to limit their use is the high Curie point of the alloys, and attempts to change the composition in order to reduce the TC usually also reduce the magnetic moment, which then negatively affects the magnetocaloric effect. There have been suggestions to use these alloys in high temperature applications [4,16], but the primary goal remains a useful magnetocaloric effect at room temperature. In this paper the focus is on substituting nickel for iron in Fe–Zr–B alloys and its effect on the magnetic properties. Not many papers have been published where they used Ni to modify the MCE of Fe-based amorphous

30

B. Podmiljsak et al. / Journal of Alloys and Compounds 591 (2014) 29–33

materials [7,17,18]. Our results show that nickel increases the magnetocaloric effect, the RC and the Curie temperature, but not in a linear fashion as is observed with other alloys [7]. 2. Experiment We studied samples with the compositions Fe84xNixZr6B10 (x = 0, 2, 4 and 6). Button ingots where prepared with an Edmund Bühler AM4 Arc Melter. These where then melt spun by inductively melting the button ingots in a partial argon atmosphere and ejecting the melt through a 0.6-mm orifice onto the perimeter of a copper wheel rotating at a surface speed of 50 m s1 with an Edmund Bühler ST1 Melt Spinner. The ribbons produced with this equipment where 2 mm wide and 25 lm thick. The crystal structure of the as-spun ribbons surface was examined by X-ray diffraction (XRD) using a Siemens D5000 diffractometer with Cu Ka radiation (2h = 5–80°). All the samples were cross-sectioned, and then polished for the JEOL JSM-7600F FEG-SEM and quantitatively assessed in terms of their composition with EDS. The magnetic properties were measured on a LakeShore vibrating-sample magnetometer (Model 7303) with a heater attached. To determine the MCE and the refrigeration capacity the magnetization was measured at discrete magnetic-field values between 0 and 1.4 T at constant temperature. This was then repeated at different temperatures, and from the magnetic measurements the DSM was calculated using an approximation of Maxwell’s relation:

DS M 

1 DT

"Z

H0

MðT þ DT; HÞdH  0

Z

#

H0

MðT; HÞdH

ð1Þ

0

3. Results and discussion The ribbons produced with the melt spinner are on the average 2 mm wide and 25 lm thick, while because of the amorphous elastic nature being endless in length. The amorphous structure of the melt-spun ribbons was also confirmed by XRD. Over a wide 2h range of 10–80° only a broad peak is seen, without any trace of sharp peaks related to a periodic lattice structure. All four specters are presented in Fig. 1. Also the images of the electron microscope showed that the ribbons are uniform with no secondary phase present. Fig. 2 shows the SEM images of ribbon x = 2. The EDS results show the same stoichiometry after the melt spinning as the weight material. This shows that there were no losses of material during sample preparation. Table 1 shows the EDS results for all four samples. Additional to the center of the ribbon we measured also the composition close to the surface. When melt spinning the surface touching the wheel cools down faster than the one exposed to argon, which can lead to a change in composition. The numbers show that this did not occur in our case. Fig. 3 shows the thermomagnetic measurements conducted on the samples. The Curie temperature for the base alloy is at 427 K. Adding nickel shifts the Cure temperature and the magnetic

Intensity (a.b.u)

x=6

x=4

x=2

x=0

0

20

40

60

Angle (2 ) Fig. 1. XRD patterns of Fe84xNixZr6B10 (x = 0, 2, 4 and 6).

80

response to higher values. For x = 2, 4 and 6 the values of the Curie temperatures are 465 K, 473 K and 482 K, respectively. The change in the magnetic entropy was calculated from the magnetic data, where we can use the Maxwell equation without any need to consider false calculations as when calculating the magnetic entropy change for a first-order magnetic transition (where the Clausius–Clapeyron equation is more suitable) [19]. The temperature dependence of the magnetic entropy change is presented in Fig. 4. While all the curves show similar shapes, we can see how large an effect nickel has on the Fe–Zr–B alloy. Nickel increases the peak temperature of the magnetic entropy change, due to an increase of the Curie temperature of the alloy. It is well known that small additions of nickel increase the TC of ferromagnetic amorphous alloys, which can be explained by the contribution of the Fe–Ni interaction (which is thought to be larger than the Fe–Fe interaction, according to the average molecular model) and the change of the non-collinear spin structures of the parent Fe–Zr–B alloy, so a more stable ‘‘ferromagnetic’’ alloy is made with its spin structure approaching a parallel alignment [20]. The Curie temperature is expected to increase up to a substitution of 50 at.% of Fe with Ni, after that it should fall quite rapidly [21]. The Fe–Ni interaction also affects the magnetic moment, which increases with increasing nickel content and is also reflected in the increase in the magnetic entropy change. Franco et al. showed in magnetocaloric materials with a SOMT that there is a ‘‘master-curve’’ behavior of the magnetic-entropychange curves not only for the same alloy at different fields, but also for different alloy compositions of the same series [22]. To obtain this master curve we have to rescale the normalized DSM(T) curves versus a temperature axis in a different way below and above the Curie temperature of the sample by ensuring that the position of two additional reference points on the curve correspond to h = ±1:

 h¼

T  T Curie =ðT r1  T Curie Þ;

T 6 T Curie

ðT  T Curie Þ=ðT r2  T Curie Þ;

T > T Curie

;

where Tr1 and Tr2 are the temperatures of the two reference points, which have been selected as those corresponding to ð1=2ÞDSpk M. Fig. 5 shows the master-curve behavior of the magnetic entropy change for the investigated alloys. It is clear that all the curves fit onto the master curve, confirming the findings in [21]. To calculate the RC from the DSM curves, different methods are used, which makes it hard to compare the values from different authors. We will calculate the RC in two different ways. In the first method [23], the RC values were obtained by numerically integrating the area under the DSM versus T curves, using the temperatures at half-maximum of the DSM peak as the integration limits (RCAREA). In the second method, we used the method of Wood and Potter (RCWP) [24], which is the most commonly used by today’s authors. They defined the refrigerant capacity for a reversible refrigeration cycle operating between Thot (the temperature of the hot reservoir) and Tcold (the temperature of the cold reservoir) as RCWP = DSMDT, where DSM is the magnetic entropy change at the hot and cold ends of the cycle (defined equal) and DT = Thot–Tcold. This represents the largest rectangle that can be inscribed inside the DSM(T) curve. The refrigerant capacity calculated using the method of wood and potter for the studied alloys is presented in Fig. 6. The graph shows the RCWP as a function of the cold reservoir temperature Tcold. The maximum in the RCWP curve represents the optimum working conditions for the refrigeration cycle. It is clear that the substitution with Ni increases the RCWP, along with DSpk M , for the samples x = 4 and x = 6. Increasing DSpk M can have also a negative effect on the RCWP [16], because the shape of the DSM(T) curve

31

B. Podmiljsak et al. / Journal of Alloys and Compounds 591 (2014) 29–33

Fig. 2. SEM images of ribbon x = 2. (a) Shows the secondary electron image of the cross section of the melt spun ribbon and (b) shows the backscattered electron image of a magnified part of figure (a) showing no additional phases being present.

Table 1 EDS results for all four samples with measurements done in the center and on the edges of the cross section of a ribbon. FNZB0

Fe Ni Zr

FNZB2

FNZB4

FNZB6

Argon side

Center

Wheel side

Argon side

Center

Wheel side

Argon side

Center

Wheel side

Argon side

Center

Wheel side

94.4

94.3

94.8

5.7

5.7

5.2

93.0 2.4 4.6

92.7 1.9 5.4

92.4 2.4 5.12

90.7 3.7 5.6

90.1 4.1 5.8

90.8 4.2 5.0

88.6 6.0 5.4

89.1 6.3 4.6

88.4 6.1 5.5

25

x=6

1

x=2 20

x=0

x=0

x=2

0.8

SM / SM(peak)

Magnetic Moment (emu/g)

x=4

15

10

5

x=4 x=6

0.6

0.4

0.2

0 340

390

440

490

0

540

-2

-1

0

1

Fig. 3. Thermomagnetic measurements for different amounts of added Ni.

Δ H=1.4T

1.2

60

1 0.8 0.6

50 40 30

0.4

20

0.2

10 400

440

480

520

x=0 x=2 x=4 x=6

70

RC (J/kg)

- Sm (J/kgK)

80

x=0 x=2 x=4 x=6

1.4

360

3

Fig. 5. Master-curve behavior of the DSM(T) curves obtained by rescaling the temperature axis for all four alloys.

1.6

0 320

2

θ

Temperature (K)

560

Temperature (K) Fig. 4. Temperature dependence of the magnetic entropy change for the alloys Fe84xNixZr6B10 (x = 0, 2, 4 and 6) for a maximum applied field of 1.4 T.

0 320

340

360

380

400

420

440

460

480

500

Tcold (K) Fig. 6. Refrigeration capacity (RCWP) of the studied alloys vs. the temperature of the cold reservoir of the refrigeration cycle for a magnetic field difference of 1.4 T.

32

B. Podmiljsak et al. / Journal of Alloys and Compounds 591 (2014) 29–33

becomes sharper, making the area inside the curve smaller. This can be seen at x = 2, where the RCWP value slightly decreases, because the positive effect of increasing DSpk M is overcome by the curve getting narrower. In Fig. 7 the dependence of T C ; DSpk M and RC on the composition is presented. A step-like increase in the TC can be observed from x = 0 to x = 2 with a monotonic increase afterwards for the samples x = 4 and x = 6. The values do not follow a power law from x = 0, as observed in previous papers [7], but they do follow it after the first addition of nickel. This suggests that the presence of nickel already has a more significant effect on the stabilization of the ferromagnetic state then the amount of it in the alloy. Also, the increase is only half the value obtained when adding the same at.% of cobalt in a similar alloy composition [5]. The DSpk M increases in a more linear way, but the real monotonic increase starts from x = 2. The values of DSpk M are 1.22 J/kg K for x = 0, 1.38 J/kg K for x = 2, 1.45 J/kg K for x = 4 and 1.52 J/kg K for x = 6. RCAREA and RCWP have a similar shape, suggesting that the DSM(T) curves have the same peak width broadness for all the investigated alloys. Both also have a minimum at x = 2, because the DSM(T) curve is getting narrower, but with more nickel the RC values increase with the highest value at x = 6, because the increase of the DSpk M overcomes the narrowing effect. RCAREA increases from 86 J/kg (x = 0) to 93 J/kg (x = 6). The values of RCWP are 59, 58, 63 and 69 J/kg for x = 0, 2, 4 and 6, respectively.

To compare these values with those reported in literature, the values have to be calculated for an applied field of 5 T. It has been shown that the field dependence of DSM, in a SOMT can be represented as DSM(T, H) = c(T)hn [25]. In our case we found that n is of the order of 0.81 for all four alloys, which we calculated from our magnetic data and is presented as the slope on the log(RC) vs. log(Field) graph in Fig. 8a. The extrapolation of the VSM magnetic measurement (with a maximum field of 1.4 T) up to 5 T using this value gives a DSpk m value of 4.26 J/kg K for x = 6. This value is smaller than that of other magnetocaloric materials, but when we make an extrapolation of RC, which is the main criterion for a good MC, we get much better results. If we use the same power law function as with DSM on RC [26], we get n = 1.16 (shown in Fig. 8b), which is close to the reported number by Caballero-Flores et al. of 1.14 [7]. RCAREA for x = 6 extrapolate to 407 J/kg for a field change of 5 T. If we compare x = 6 with Gd5Ge1.9Si2Fe0.1, which has a RCAREA value of 355 J/kg for DH = 5 T [7], we can see that the value is higher. Also, the temperature span for the x = 6 alloy is DT = Thot–Tcold  105 K, which is close to the 90 K of Gd5Ge1.9Si2Fe0.1. Comparable Fe-based amorphous alloys have higher RC values, but the temperature spans for these values are much higher [7,21] and calculating them for a span of 90 K reduces these values below our results. Even if the working temperature of our material is too high, it has good properties and shows that a lot of modifications can be made to achieve better values. Furthermore, the cost of these alloys is about 15 times lower than that of Gd-based alloys.

480 470

TC (K)

460

(a) 0.2

450 0

440

log ( S)

430 420

-0.2 -0.4

- S M(pk) (J/kgK)

1.5 -0.6 1.4

-0.8 -0.6

-0.2

0

0.2

log (Field)

(b)

1.2 90

2.5 2

80

RC (AREA) RC (WP)

70 60

log (RC)

RC (J/kg)

-0.4

1.3

1.5 1 0.5

50 0

2

4

6

Ni content x (at.%)

0 -0.6

-0.4

-0.2

0

0.2

log (Field) Fig. 7. The influence of Ni content on T C ; DSpk M and RC of the amorphous Fe84xNixZr6B10 (x = 0, 2, 4 and 6) alloy series. An applied field of 1.4 T was used pk to measure DSM and RC.

Fig. 8. Field dependence of the (a) DSpk M and of (b) RCAREA calculated from the magnetic data, with the calculated n for both slops.

B. Podmiljsak et al. / Journal of Alloys and Compounds 591 (2014) 29–33

4. Summary and conclusion In summary, it has been show that Ni increases T C ; DSpk M and at higher additions also RC, which can be attributed to the Fe–Ni interaction, which is thought to be larger than the Fe–Fe interaction. The TC should increase up to x  42, where the Fe:Ni content is at 50:50 and the Fe–Ni interaction is at its peak. Whether the magnetocaloric effect has a similar behavior needs to be investigated. Also, the magnetic entropy change for all the presented alloys overlaps with the same master-curve behavior. The RC values are higher compared to Gd5Ge1.9Si2Fe0.1, with the same temperature span, making these alloys, additional to the lower material costs, even more favorable as a working material for magnetic refrigerators. Acknowledgments This work has been supported by the Ministry of Higher Education, Science and Technology of the Republic of Slovenia (P2-0084) and by the NAMASTE Centre of Excellence. References [1] P. Duwez, C.H. Lin, J. Appl. Phys. 38 (1967) 4096. [2] T.D. Shen, R.B. Schwarz, J.Y. Coulter, J.D. Thompson, J. Appl. Phys. 91 (2002) 5240.

33

[3] I. Škorvánek, J. Kovacˇ, Czech. J. Phys. 54 (2004) 189. [4] V. Franco, J.S. Blázquez, C.F. Conde, A. Conde, Appl. Phys. Lett. 88 (2006) 042505. [5] V. Franco, J.S. Blázquez, A. Conde, J. Appl. Phys. 100 (2006) 064307. [6] J. Du, Q. Zheng, Y.B. Li, Q. Zhang, D. Li, Z.D. Zhang, J. Appl. Phys. 103 (2008) 023918. [7] R. Caballero-Flores, V. Franco, A. Conde, K.E. Knipling, M.A. Willard, Appl. Phys. Lett. 96 (2010) 182506. [8] E. Warburg, Ann. Phys. Chem. 13 (1881) 141. [9] V.K. Pecharsky, K.A. Gschneidner Jr., Phys. Rev. Lett. 78 (1997) 4494. [10] F.X. Hu, B.G. Shen, J.R. Sun, Z.H. Cheng, G.H. Rao, X.X. Zhang, Appl. Phys. Lett. 78 (2001) 3675. [11] O. Tegus, E. Brück, K.H.J. Buschow, F.R. De Boer, Nature 415 (2002) 150. [12] F.X. Hu, B.G. Shen, J.R. Sun, G.H. Wu, Phys. Rev. B 64 (2001) 132412. [13] K.A. Gschneidner Jr., V.K. Pecharsky, Annu. Rev. Mater. Sci. 30 (2000) 387. [14] V. Provenzano, A.J. Shapiro, R.D. Shull, Nature (London) 429 (2004) 853. [15] L. Morellon, P.A. Algarabel, M.R. Ibarra, J. Blasco, B. García-Landa, Z. Arnold, F. Albertini, Phys. Rev. B 58 (1998) R14721. [16] F. Johnson, R.D. Shull, J. Appl. Phys. 99 (2006) 08K909. [17] J.J. Ipus, H. Ucar, M.E. McHenry, IEEE Trans. Magn. 47 (2011) 10. [18] N.H. Dan, N.H. Duc, T.D. Thanh, N.H. Yen, P.T. Thanh, N.A. Bang, D.T.K. Anh, T.-L. Phan, S.-C. Yu, J. Korean Phys. Soc. 62 (2013) 1715. [19] G.J. Liu, J.R. Sun, J. Shen, B. Gao, H.W. Zhang, et al., Appl. Phys. Lett. 90 (2007) 032507. [20] S. Bao-gen, X. Rufeng, Z. Jiang-gao, Z. Wen-shan, Phys. Rev. B 43 (1991) 11005. [21] S.A. Kostyrya, B. Idzikowski, J. Magn. Magn. Mat. 304 (2006) e537. [22] V. Franco, C.F. Conde, A. Conde, L.F. Kiss, Appl. Phys. Lett. 90 (2007) 052509. [23] K.A. Gschneidner Jr., V.K. Pecharsky, A.O. Pecharsky, C.B. Zimm, Mater. Sci. Forum 315 (1999) 69. [24] M.E. Wood, W.H. Potter, Cryogenics 25 (1985) 667. [25] V. Franco, A. Conde, Int. J. Refrig. 33 (2010) 465. [26] V. Franco, A. Conde, J.M. Romero-Enrique, J.S. Blázquez, J. Phys.: Condens. Matter 20 (2008) 285207.