Corrosion Behavior of Commercial Magnetic Refrigerant Gadolinium in Water

JOURNAL OF RARE EARTHS .Vol. 22, Spec. , Dec. 2004, p. 099Corrosion Behavior of Commercial Magnetic Refrigerant Gadolinium inWaterZhang Zeyu(张泽玉)*, Long Yi(龙毅), Wen Da(闻达), Ye Rongchang(叶荣昌), Wan Farong(万发荣)(School of Materials Science and Engineering, University of Science and Technology Bejing, Bejing 100083,China)Abstract: Gadolinium( Gd ) used as magnetic refrigerant always works in water environment. However, its poor COrTO-sion resistance is serious impediment against wider application of Gd. In this paper, the corrosion behavior of two typesof commercial Gd (A, B both are 98.9 at. % pure) with the same oxygen content has been studied. The results showthat the corrosion rate of A is 3. 226 times higher than that of B in deionized water and 6. 039 times in tap water. Ac-cording to SEM, the different corrosion rate is because of the different ditribution of impurity in matrix. In addition ,NaOH solution was chosen as inhibitor to prevent Gd from being corroded scessfully. No pitting corrosion and weightloss were observed for commercial Gd even after immersion for nearly 2000 h in NaOH solution.Key words: gadolinium; corrosion behavior ; impurity; pitting corrosionCLC number: TM273Document code: AAricle ID: 1002 -0721 (2004) -0099 -04Gadolinium is a very important magnetic refrig-1 Experimentalerant in near- room temperature magnetic refrigerationthat is a new environment-safe refrigeration technolo-Two commercial Gd samples A and B are bothgy and a promising alternative to conventional vapor-98.9 at. % pure. The major impurities of samplescycle refrigeration near room temperature. The in-are interstitial elements (in A-0. 80，<0. 10 C,fluence of impurity on magnetocaloric effect ( MCE)0.16N, 0.08 Ca; in B-0.8 O, 0.08 C, 0.16 N,of ,Gd has been studied. According to Zimm, the<0.04 Ca at. % ). The preparations of samples areMCE of commercial Gd (93 at% pure) was ~ 5%described as follows. In order to obtain gadoliniumlarger below 285 K，~ 5% smaller above 300 K ,fluorides as the starting salt, anhydrous HF gas wasand smaller at 293 K than high purity Gd (99.9 at%directly over gadolinium oxide at elevated tempera-pure)2). The presence of large amounts of impuritytures. Then, two methods were employed to preparein commercial Gd lowered the second- order paramag-Gd metal. For A, calcium and gadolinium fluoridenetic-ferromagnetic transition temperature， andwere mixed together and heated; excess quantity ofcaused some erroneous results in MCE determined incalcium was used above that required by the stoichi-pulsed magnetic fields3. However , commercial Gdometry. The metal was melt again with gadoliniumis still used in the magnetic refrigerator for the largefluoride to eliminate the surplus calcium. For B, thecost difference between commercial and high purityprocess was based on the formation of an intermedi-Gds- 10。 In most magnetic refrigerators, Gd was im-mersed in water medium that was chosen as heatate Gd-Mg alloy. The charge, consisting of a mix-ture of GdF3, CaCl2，granular calcium and magnesi-transfer fluid. Therefore, it is important to study them metals, was heated in a reaction crucible. Thecorrosion behavior of commercial Gd in water. How-ever, the corrosion behaviors of commercial Gd inGd-Mg alloy was then heated in an evacuated systemwater have been scarcely reported. In this paper, theto remove magnesium and calcium. The resutingcorrosion behavior and corrosion resistance of com-sponge was arc-melted into ingot form.mercial Gd with different impurity distribution haveThe corrosion behaviors of sample A and B werebeen presented.eval中国煤化工. measurements anweigsurements were per-MHCNMHGReceived date: 2004 -08 - 10, revised date: 2004 - 10 -20Foundation item: Project spprted by "863" program (2002AA324010) and National Natural Science Foundation of China(50071006)Biography: Zhang Zeyu( 1976- ), Male, DoctorCorresponding author (E - mail: zeyu_ zhang@ 126. com)100JOURNAL OF RARE EARTHS, Vol.22, Spec. , Dec. 20041.0(ab)0.5BI0.0而-0.5-1.0--1.5Fig.2 Photographs of(a)A and ( b)B immersed2.0-in deionized water for 168 h-9-8-7-6-5-4-3-2log()JA metal8B metalFig.1 Potentiodynamic polarization curves豆1.2of A and B in deionized water票1.0formed in a three-elctrode cell using a platinum0.8|mesh as a counter electrode and a saturated0.6|calomel electrode ( SCE) as a reference elec-0.4trode. Potentiodynamic polarization curves were0.2measured with a potential sweep rate of 18 mV .min'0.0LTap waterDeionized waterto construct the Tafel plots ( logarithmic variation ofcurrent as a function of voltage). The working elec-trode was made by attaching Cu wire to one end ofFig.3 Corrosion rates of A and B in dif-the specimens and sealing into a plastic tube with ep-ferent wateroxy resin. Before immersion test for the detection 0To further investigate the different corrosion re-weight loss, the specimens were carefully groundsults for A and B with nearly same purity, the impu-with silicon carbide papers up to NO. 1000， de-rity distribution and corrosion behaviors were ana-greased in acetone and dried in air. After immersion ,lyzed by optical microscopy and SEM. Fig. 4 ilus-the specimens were cleaned manually. The surface oftrates the metallographs of A and B that have notspecimen was ground again and new fresh solutionbeen immersed in water. There are many circular andwas used for each measurement. Magnetization werelttesized impurities, which are dispersed uniformlymeasured from 278 K to 310 K by the superconduct-in sample A. The sample B had fewer impurities ，ing quanta interfere device ( SQUID).which are bigger than that in sample A but dispersed2 Results and Discussionmore concentratedly. The SEM of sample A and Bthat immersed in deionized water for 1 hour arePotentiodynamic polarization curves for A and Bshown in Fig. 5. It is obvious that corrosion occurredimmersed in deionized water at room temperature areat the boundary between the impurity and matrix forshown in Fig. 1. The free corrosion potential ( Eorn)sample A and at the grain boundary for sample B.of A andB are -0.925 V and -0. 672 V. The corro-This indicates that the dispersion of impurity affectssion current density (Iom) ofA and B arethe corrosion resistance for the formation of corrosion1.95 μA .cm-2 and0.81 μA .cm~. It can be seencell.that sample B may have a better corrosion resistanceMagnetic entropy change of sample A is shownthan sample A.in Fig. 6, which was calculated by Eq. (1) that de-The immersion results shown in Fig. 2 confirmduced from Maxwell relation.that A is more likely corroded than B. From the fig-ure, A was corroded more seriously than B and bothwere covered with loose corrosion products. After e中国煤化工rasing the products, there were pitting corrosions 0C-curring on the surfaces of specimens. From theMHCNMHGweight loss of A and B immersed in tap water and0 05mmdeionized water shown in Fig. 3, the corrosion rate ofA is 3.226 times higher than that of sample B in tapFig. 4 Metallographs of(a)A and(b)Bwater and 6. 039 times higher in deionized water.ZhangZY et al .Corrosion Behavior of Magnetic Refrigerant Gadolinium101(b)grain boundaryFig.5 SEM of A(a) and B( b) immersed in deionized water for1 hcesfully chosen NaOH solution as an excellent inhib-10-itor for Gd. The polarization curves of A and B inNaOH solution (pH = 10) show that A and B are8spontaneously passivated in NaOH solution in a wideregion (Fig. 7).Table 1 lists the two parameters， Ecorr and2T-s 4{Icorr，calculated from Figs. 7 and 1. It can be con-cluded from the potentiodynamic polarization curves1.5T“2that the addition of NaOH solution inhibited the a-280290300ionodic reaction of both A and B greatly and decreasedT/Kthe Icorr to about one order. The Ecorr is also driftedFig.6 Magnetic entropy changes for metal A at4,2 ,5Tto values that are more positive.Table 1 Ecorr and Icorr calculated from polarization1.5---o-Bcurves for A and B in NaOH solution1.0_AB0.5Ee百0.0(pAVam)Deionized water -0.925 0.23-0.672 0.81-0.5pH=10 .-0.5901.95-0.588 0.11-1.0-1.5-8-7-6-5-4-3Fig. 8 is the photograph of A immersed in NaOHsolntion(pH = 10) for 1960 h. The surface of A islog(i)still brightness and silver white. Pitting corrosion andFig. 7 polarization curves of A and B in NaOH solution( pH 10)weight loss were not observed in this process. TheAS,=严aM(H.T)dH(1)same result was with B. There are obvious passiva-JHtion for A and B in NaOH solution (pH=10) in de-where SM is the magnetic entropy, H the magneticspite of the difference of amount and distribution offield, M the magnetization and T the absolute tem-impurity.perature. The maximum entropy changes are respec-tively4.1 J.kg~'.K-for a field change of 1.5 Tand9.4 J . kg~I●K~' for a field change of 5 T.These values are agreement with the referencsll.21.Therefore，the distribution of impurity has much中国煤化工effct on the corrosion resistance but lttle on its mag-netic entropy change.MHCNMHGThe results shown above indicate the corrosionof Gd even in deionized water is very serious. There-fore, it is necessary to find an inhibitor to ensure theFig.8 Photograph of A immersed in NaOH solutionstability of Gd in water environments. We have suc-(pH=10) for 1960 h102JOURNAL OF RARE EARTHS, Vol.22, Spec. , Dec. 2004effet and magnetic refrigeration [J]. J. Magn. Magn.3 ConclusionMater, 1999, 200: 44.The weight loss results show that the corrosion[4] Brown G V. Magnetic heat pumping near room tem-perature [J]. J. Appl. Phys., 1976, 47: 3673.rate of A is 3. 226 times higher than that of B in dei-[5] Kirol L D, Dacus M W. Rotary recuperative magneticonized water and 6.039 times in tap water. The cor-heat pump[J]. Adv. Cryog. Eng., 1987, 757.rosion current density (Lom) of A and B are 1.95[6] Steyert Jr. High temperature rfrigerator [P]. U. S.μA/cm2 and 0. 81 μA/cm2 in deionized water. TheA.: U. s. Patent4,107 ,935, 1978. .metallographs and SEM of sample A and B indicate[7] Howard. Magnetic heat pump flow direetor [P]. U.that the distribution of impurity plays an importantS.A.: U.S. Patent5 ,444 ,983, 1995.8] Ziolo. Magnetic refrigerant compositions and processesrole in the corrosion resistance of commercial Gd.for making and using[P]. U.S.A.: U. s. Patent 5,Otherwise, no pitting corrosion and weight loss641 ,424, 1997.are recognized for sample A and B immersed in[9] Gschneidner K A. Active magnetic refrigerants basedNaOH solution (pH=10) for 1960 h. The result ofon Gd-Si-Ge material and refrigeration apparatus andweight loss measurement and polarization curves indi-process[P]. U.S.A. : U. S. Patent 5 ,743 ,095, 1998.cates that NaOH solution could inhibit the corrosion[10] Lawton Jr.， et al. Reciprocating active magnetic re-of Gd in deionized water.generator refrigeration apparatus [P] . U.S.A.: U.S. Patent5 ,934 ,078, 1999.References :[11] GuoZB, Du Y w, ZhuJ s et al. Large magnetic en-tropy change in perovskit-type manganese oxides [J].[1] Glanz J. Making a bigger chill with magnets [J]. Sci-Phys Rev. Lett, 1997, 78: 1142.ence, 1998, 279: 2045.[12] Pecharsky V K, Gschneidner K A. Giant magnetoca-[2] Zimm C B, Jatrab A, Stermberg A, et al. Descriptionloric efete in Gd, (Si,Ge.)[J]. Phys. Rev. Lelt. ,and preformance of a near-room temperature magnetic1997, 78: 4494.refrigerator [J]. Adv. Cryog. Eng. , 1998, 43: 1759.[3] Pecharsky V K, Gschneidner K A. Magnetocaloric中国煤化工MYHCNMHG