EPJ manuscript No. (will be inserted by the editor) 4 0 0 2 Comparison among the local atomic order of amorphous TM-Ti n alloys (TM=Co, Ni, Cu) produced by Mechanical Alloying a J studied by EXAFS 1 3 K. D. Machado, J. C. de Lima, C. E. M. Campos, and T. A. Grandi ] Departamento deF´ısica, UniversidadeFederal deSanta Catarina, 88040-900 Florian´opolis, SC, Brazil i c s - the dateof receipt and acceptance should beinserted later l r t Abstract. We have investigated the local atomic structure of amorphous TM-Ti alloys (TM = Co, Ni, m Cu)producedbyMechanicalAlloyingbymeansofEXAFSanalysesonTMandTiK-edges.Coordination t. numbersandinteratomicdistancesforthefouralloyswherefoundandcompared.EXAFSresultsobtained a indicatedashorteningintheunlikepairsTM-TiasthedifferencebetweendelectronsofTMandTiatoms m increases, suggesting an increase in thechemical short range order (CSRO)from TM = Co to Cu. - d Key words. amorphous alloys – EXAFS – mechanical alloying n o PACS. 61.43.Dq amorphous alloys – 61.10.Ht EXAFS– 81.20.Ev mechanical alloying c [ 1 1 Introduction and high sensitivity to the chemical environment around v a specific type of atom of an alloy, EXAFS is a technique 3 Mechanicalalloying(MA)technique[1]isanefficientmethod(good reviews are found in Refs. [18,19,20,21]) very suit- 1 for synthesizing crystalline [2,3,4,5] and amorphous [6,7, able to investigate the local atomic order of crystalline 0 2 8,9,10] materials, as well as stable and metastable solid compounds andamorphousalloys.Anomalouswide angle 0 solutions[11,12].MAhasalsobeenusedtoproducemate- x-ray scattering (AWAXS) is also a selective technique, 4 rialswithnanometersizedgrainsandalloyswhosecompo- but due to the small Kmax that can be achievedon Ti K- 0 nents havelargedifferences in their melting temperatures edge(∼4˚A−1),little informationcouldbeobtainedfrom / and are thus difficult to produce using techniques based an AWAXS experiment at the Ti K-edge. On the other t a on melting. It is a dry milling process in which a metal- hand, EXAFS measurements performed on this edge ex- m lic or non-metallic powder mixture is actively deformed tendedupto∼16˚A−1 insomealloys,allowingthedeter- - in a controlled atmosphere under a highly energetic ball minationofstructuraldatawith reasonableaccuracy.Us- d charge.The few thermodynamics restrictionsonthe alloy ingthistechnique,wehavedeterminedcoordinationnum- n compositionopenupawiderangeofpossibilitiesforprop- bers and interatomic distances in the first coordination o c erty combinations [13], even for immiscible elements [14]. shell of a-Cu64Ti36. In addition, the chemical short range : The temperatures reached in MA are very low, and thus order (CSRO) in the three TM-Ti alloys were compared v thislowtemperatureprocessreducesreactionkinetics,al- anditincreasesasTMgoesfromCotoCu,inaccordance i X lowingtheproductionofpoorlycrystallizedoramorphous withresultsreportedbyHausleitnerandHafnerusingMD r materials. simulations [22]. a WehaveusedMAtoproducethreeamorphousTM-Ti alloys: Co57Ti43 (a-Co57Ti43), Ni60Ti40 (a-Ni60Ti40) and Cu64Ti36 (a-Cu64Ti36) starting from the crystalline ele- 2 Experimental Procedure mental powders. In Ref. [6] a-Co57Ti43 was studied by EXAFS and x-ray diffraction. In Ref. [7] we found coor- BlendedTM(TM=Co,NiandCu)andTicrystallineel- dination numbers and interatomic distances for the first ementalpowders(Co:Vetec,99.7%,particlesize<10µm; neighbors for a-Ni60Ti40 using EXAFS and RMC simula- Ni:Merck,99.5%,particlesize<10µm;Cu:Vetec99.5%, tions [15,16,17] of its total structure factor S(K). Here, particle size <10 µm; Ti: BDH, 99.5%,particle size <10 westudieda-Cu64Ti36 byEXAFS,andcoordinationnum- µm), with initial nominal compositions TM60Ti40, were bers and interatomic distances were found and compared sealedtogetherwithseveralsteelballs,underanargonat- to those found for the other alloys. Due to its selectivity mosphere,inasteelvial(moredetailscanbeseenatRefs. Send offprint requests to: K. D.Machado [6]and[7]).Theball-to-powderweightratiowas5:1forthe Correspondence to: kleber@fisica.ufsc.br four alloys. The vial was mounted in a Spex 8000 shaker 2 K.D. Machado et al.: Comparison among thelocal atomic order of amorphous TM-Ti alloys studied by EXAFS mill and milled for 9 h. A forced ventilation system was 0.10 used to keep the vial temperature close to room temper- (a) ature. The composition of the as-milled powder was mea- sured using the Energy Dispersive Spectroscopy (EDS) ] technique,givingthecompositionsCo57Ti43,Ni60Ti40and k) 0.05 ( Cu64Ti36,andimpuritytraceswerenotobserved.EXAFS 3c k measurements were carriedout on the D04B beam line of [ F LNLS(Campinas,Brazil),using achannelcutmonochro- T mator (Si 111), two ionization chambers as detectors and 0.00 a 1 mm entrance slit. This yielded a resolution of about 1.6 eV at the Ti K edge and 3.9 eV at the Co, Ni and Cu 0.03 (b) K edges. All data were taken at room temperature in the transmissionmode.Theenergyandaveragecurrentofthe )] 0.02 storage ring were 1.37 GeV and 120 mA, respectively. k ( 3c k 0.01 [ F T 3 Results and Discussion 0.00 The EXAFS oscillationsχ(k) on Cu and Ti K edges of a- 0 2 4 6 8 Cu64Ti36 are shown in Fig. 1 weighted by k3. After stan- k (Å-1) dard data reduction procedures using Winxas97 software [23], they were filteredby Fourier transformingk3χ(k)on Fig. 2. Fourier transformation of experimental EXAFS spec- both edges (Cu edge, 3.25 – 13.6 ˚A−1 and Ti edge, 3.5 – tra: a) at the Cu K-edgeand b) at theTi K-edge. 14.5˚A−1)usingaHanningweightingfunctionintor-space (Fig.2)andtransformingbackthefirstcoordinationshells Figure3showstheexperimentalandthefittingresults (1.30– 2.67˚A for Coedge and1.85–3.24˚Afor Tiedge). for the Fourier-filtered first shells on Cu and Ti edges. Filtered spectra were then fit by using Gaussiandistribu- Structural parameters extracted from the fits, including tions to represent the homopolar and heteropolar bonds the errors in these values, are listed in table 1. This table [24].We alsousedthe thirdcumulantoptionofWinxas97 also shows data concerning crystalline Cu2Ti (c-Cu2Ti, to investigate the presence of asymmetric shells. The am- JCPDScardNo 200371).ItcanbeseenfromTable1that plitude and phase shifts relative to the homopolar and the number of Cu-Cu pairs in a-Cu64Ti36 is much higher heteropolar bonds needed to fit them were obtained from than it is in c-Cu2Ti, whereas the Cu-Cu average inter- ab initio calculations using the spherical waves method atomicdistanceincreasesintheamorphousalloy.Thereis [25] and FEFF software. a shortening in the Cu-Ti distance in a-Cu64Ti36, which becomesshorterthattheCu-Cuaveragedistance,butthe number of these pairs are almost the same. Concerning Ti-Ti pairs, there is a reduction both in its number and 3 (a) distance when compared to c-Cu2Ti. As it can be seen in Table 1, on the Cu edge, two Cu-Cu subshells and one Cu-Ti shell were considered in ) 0 (k the first shell in order to find a good fit, whereas on the 3c Ti edge one shell of Ti-Cu and one shell of Ti-Ti were k -3 needed. We started the fitting procedure using one shell for all pairs. This choice did not produce a good quality fit of Fourier-filtered first shells on the Cu edge. Besides 1.5 that, the well-known relations below were not verified in (b) fitting EXAFS data: ) 0.0 k c N =c N ( i ij j ij 3c r =r k ij ji σ =σ -1.5 ij ji where c is the concentration of atoms of type i, N is 4 6 8 10 12 14 i ij the number of j atoms located at a distance r around ij k (Å-1) an i atom and σij is the half-width of the Gaussian. By considering two Cu-Cu sub-shells the quality of the fit on Fig. 1. WeightedexperimentalEXAFSspectra:(a)attheCu the Cu edge was much improved (see Fig. 3). Moreover, K edge and (b) at the Ti K edge. the relations above were satisfied. K.D. Machado et al.: Comparison among thelocal atomic order of amorphous TM-Ti alloys studied by EXAFS 3 Figure5showsthe threeweightedEXAFSoscillations 4 (a) k3χ(k) on the Ti K edges. k3χTi(k) for a-Co57Ti43 and a-Ni60Ti40 are very similar, indicating that Ti atoms are 2 foundinsimilarchemicalenvironments.However,k3χTi(k) ) (k for a-Cu64Ti36 shows a different behavior, which reflects 0 3ck the differences found in the coordination number deter- minedforTiatoms inthis alloy.Fora better comparison, -2 structuraldatafound forthe otheralloysarelistedinTa- ble 2. -4 (b) 1 ) 0 (k (c) 3c k -1 ) k ( 3c k 3 6 9 12 15 k (Å-1) (b) Fig. 3. Fourier-filtered first shell (full line) and its simulation (squares) for a-Cu64Ti36 at the(a) Cu K edge, (b)Ti Kedge. (a) 4 6 8 10 12 14 16 -1 k (Å ) (c) Fig.5. WeightedEXAFSoscillationsk3χ(k)ontheTiKedge for (a) a-Cu64Ti36,(b) a-Ni60Ti40 and (c) a-Co57Ti43. ) k ( From Tables 1 and 2, it can be seen that the number 3ck of TM-TM pairs increases from TM = Co to Cu, and the same behavior is seen in the interatomic distances. Con- (b) cerning Ti-Ti pairs, there is an increase in the number of these pairs from Co to Ni, but in a-Cu64Ti36 it decreases byalmost3atoms,indicatingadifferentchemicalenviron- (a) mentaroundTiatomsinthisalloy.Thisfactisreinforced if Ti-Ti interatomic distances are considered. There is a small reduction in the distance from Co to Ni and a very large reduction from Ni to Cu. It should be noted that 4 6 8 10 12 14 16 theaverageCu-CuinteratomicdistancefoundbyEXAFS -1 analysis is larger than that for Cu-Ti pairs. This is an k (Å ) important feature, and it is also seen in a-Ni60Ti40. In Fig. 4. Weighted EXAFS oscillations k3χ(k) on the TM K fact, considering TM-Ti pairs, it can be seen that their edge for (a) a-Cu64Ti36, (b) a-Ni60Ti40 and (c) a-Co57Ti43. interatomic distances decrease with the TM atomic num- ber. The number of TM-Ti pairs also lowers when TM goesfromCotoCu.AccordingtoHausleitnerandHafner [22], for TM-Tialloys this shortening in the TM-Ti inter- It is interesting now to compare EXAFS data found atomic distance is associatedwith a changeof the d-band for the three alloys. Fig. 4 shows the weighted EXAFS electronic density of states in the TM-Ti alloys from a oscillations k3χ(k) obtained on the TM K edges of the common-bandtoasplit-bandform,whichdependsonthe three TM-Tialloys.The oscillationson the three K edges difference of d-electrons in the TM and Ti atoms. Since are very similar, with small differences in the amplitudes this difference increases from Co to Cu, we should expect of the oscillations. thatthe shorteningeffect would be weakerin a-Co57Ti43, 4 K.D. Machado et al.: Comparison among thelocal atomic order of amorphous TM-Ti alloys studied by EXAFS increasingfora-Ni60Ti40 andreachingitsmaximumvalue 13. J. M. Poole, J. J. Fischer, Mater. Technol. 9 (1994) 21. ina-Cu64Ti36,andthisisverifiedinourresults.Fukunaga 14. M.Abbate,W.H.Schreiner,T.A.Grandi,J.C.deLima, etal.alsofoundthisshorteningina-Ni40Ti60 producedby J. Phys.:Cond. Matter 13 (2001) 5723. MQ [26]. It is important to note that this shortening ef- 15. R.L.McGreevy,L.Pusztai,Mol.Simulations1(1988)359. fect was also confirmed by RMC simulations [15,16,17] 16. RMCA version 3, R.L. McGreevy, M. A.Howe and J. D. in a-Ni60Ti40 [7] and a-Cu64Ti36 (results to be published Wicks, 1993. available at http://www.studsvik.uu.se. elsewhere). This reduction in the TM-Ti interatomic dis- 17. R.L.McGreevy,J.Phys.:Condens.Matter13(46)(2001) tance indicates that the CSRO in TM-Ti alloys increases 877. 18. P.A.Lee,P.Citrin,P.Eisenberger,B.Kincaid,Rev.Mod. from Co-Ti to Cu-Ti alloys. Phys. 53 (1981) 769. 19. B. K. Teo, D. C. Joy, EXAFS Spectroscopy, Techniques and Applications, Plenum, New York,1981. 4 Conclusion 20. T. M. Hayes, J. B. Boyce, Solid State Physics, Vol. 37, Academic Press, New York,1982, p.173. An amorphous Cu64Ti36 alloy was produced by Mechani- 21. D.C.Koningsberger,R.Prins,X-rayAabsorption,Wiley, cal Alloying technique and its local atomic structure was New York,1988. determined from EXAFS analysis. We could find coor- 22. C.Hausleitner,J.Hafner,Phys.Rev.B45(1)(1992)128. dination numbers and interatomic distances for the first 23. T. Ressler, J. Phys.7 (1997) C2. neighbors of this alloy. The results obtained were com- 24. E. A. Stern, D. E. Sayers, F. W. Lytle, Phys. Rev. B 11 pared to those found in c-Cu2Ti and also with those de- (1975) 4836. termined for other two TM-Ti alloys (TM = Co and Ni) 25. J. J. Rehr,J. Am.Chem. Soc. 113 (1991) 5135. producedbyMA.Themostimportantfeatureconsidering 26. T. Fukunaga, N. Watanabe, K. Suzuki, J. Non-Cryst. thesealloysisthedecreaseinTM-Tiinteratomicdistance Solids 61 & 62 (1984) 343. as a function of the TM atomic number, as proposed by Hausleitner and Hafner [22] to TM-Ti alloys. This effect can be associated to the CSRO in the alloys, which in- creases from Co to Cu. We thank the Brazilian agencies CNPq, CAPES and FINEP forfinancialsupport.WealsothankLNLSstaffforhelpduring measurements (proposal nos XAS 799/01 and XAS 998/01). This study was partially supported by LNLS. References 1. C. Suryanarayana,Prog. Mater. Sci. 46 (2001) 1. 2. K. D. Machado, J. C. de Lima, C. E. M. Campos, T. Grandi, A. A. M. Gasperini, Sol. State Commun. 127 (2003) 477. 3. C. E. M. Campos, J. C. de Lima, T. A. Grandi, K. D. Machado, P. S. Pizani, Sol. State Commun. 123 (2002) 179. 4. C. E. M. Campos, J. C. de Lima, T. A. Grandi, K. D. Machado, P. S. Pizani, Physica B 324 (2002) 409. 5. J.C.D.Lima,E.C.Borba,C.Paduani,V.H.F.dosSan- tos, T. A. Grandi, H. R. Rechenberg, I. Denicol´o, M. El- massalami, A.P.Barbosa, J.AlloysComp.234(1996) 43. 6. K.D.Machado,J.C.deLima,C.E.M.deCampos,T.A. Grandi,A.A.M.Gasperini, Sol.StateCommun.Submit- ted to publication. 7. K.D.Machado,J.C.deLima,C.E.M.deCampos,T.A. Grandi, D.M. Trichˆes, Phys. Rev.B 66 (2002) 094205. 8. J.C.D.Lima,D.M.Trichˆes,T.A.Grandi,R.S.deBiasi, J. Non-Cryst.Solids 304 (2002) 174. 9. A.W. Weeber, H.Bakker, Physica B 153 (1988) 93. 10. C.E.M.Campos,J.C.deLima,T.Grandi,K.Machado, P.Pizani, Sol. StateCommun. 126 (2003) 611. 11. D. K. Mukhopadhyay, C. Suryanarayana, F. H. Froes, ScriptaMetall. Mater. 30 (1994) 133. 12. A. R. Yavari, P. J. Desr´e, T. Benameur, Phys. Rev. Lett. 68 (1992) 2235. K.D. Machado et al.: Comparison among thelocal atomic order of amorphous TM-Ti alloys studied by EXAFS 5 Table 1. Structural data determined for a-Cu64Ti36. The numbersin parenthesis are theerrors in thevalues. EXAFS Cu K-edge Ti K-edge R factor 4.2 5.4 Bond Type Cu-Cu1 Cu-Ti Ti-Cu Ti-Ti N 2.5 (0.4) 7.0 (1.0) 4.4 (0.6) 7.8 (1.0) 2.8 (0.4) r (˚A) 2.31 (0.02) 2.80 (0.01) 2.42 (0.02) 2.42 (0.02) 2.70 (0.01) σ2 (˚A×10−2) 0.925 (0.1) 2.51 (0.2) 3.47 (0.4) 3.47 (0.4) 1.01 (0.1) c-Cu2Ti Bond Type Cu-Cu2 Cu-Ti3 Ti-Cu4 Ti-Ti5 N 2 2 2 2 1 4 4 2 2 2 r (˚A) 2.54 2.58 2.58 2.61 2.63 2.58 2.61 2.63 2.93 3.32 1There are 9.5 pairs at hri=2.67 ˚A. 2There are 4 pairs at hri=2.56 ˚A. 3There are 5 pairs at hri=2.60 ˚A. 4There are 10 pairs at hri=2.60 ˚A. 5There are 4 pairs at hri=3.08 ˚A. Table 2. Structuraldata found for amorphous TM-Ti alloys (averagevalues).Thenumbersinparenthesis aretheerrors in thevalues. a-Co57Ti43 [6] Co K-edge Ti K-edge R factor 1.9 2.9 Bond Type Co-Co Co-Ti Ti-Co Ti-Ti N 6.0 (0.8) 6.0 (0.8) 7.9 (1.0) 4.9 (0.7) r (˚A) 2.50 (0.01) 2.52 (0.01) 2.52 (0.01) 2.98 (0.02) σ2 (˚A×10−2) 1.45 (0.2) 4.57 (0.6) 4.57 (0.6) 1.36 (0.1) a-Ni60Ti40 [7] Ni K-edge Ti K-edge R factor 1.5 3.5 Bond Type Ni-Ni Ni-Ti Ti-Ni Ti-Ti N 8.8 (1.2) 5.2 (0.6) 7.9 (1.0) 5.5 (0.8) r (˚A) 2.58 (0.02) 2.55 (0.01) 2.55 (0.01) 2.93 (0.02) σ2 (˚A×10−2) 2.03 (0.3) 0.33 (0.04) 0.33 (0.04) 0.003 (0.0008)