UDK 669:538.945:536 ISSN 1580-2949 Original scientific article/Izvirni znanstveni članek MTAEC9, 44(1)3(2010) THE THERMAL CONDUCTIVITY OF Al73Mn27 ^Fe^ TAYLOR PHASES TOPLOTNA PREVODNOST TAYLORJEVIH FAZ Al73Mn27-xFex Denis Stanič1,2, Petar Popcevic1, Igor Smiljanic1, Zeljko Bihar1, Jagoda Lukatela1, Boran Leontic1, Ante Bilušic1,3, Ivo Batistic4, Ana Smontara1 1Laboratory for the Study of Transport Problems, Institute of Physics, Bijenička 46, POB 304, HR-10000 Zagreb, Croatia 2Department of Physics, University of Osijek, Gajev trg 6, HR-31000 Osijek, Croatia 3Faculty Science, University of Zagreb, Bijenička c. 32, HR-10000 Zagreb, Croatia 4Department of Physics, Faculty of Natural Sciences of the University of Split, Nikole Tesle 12, HR-21000 Split, Croatia dstanic@fizika.unios.hr Prejem rokopisa - received: 2009-07-21; sprejem za objavo - accepted for publication: 2009-08-24 The thermal conductivity (k) of Al73Mn27-jFe;, (x = 0, 2, 4, 6) complex metallic alloys has been measured in the temperature interval from 2 K to 300 K. All the alloys are Taylor (T) phases, except Al73Mn21Fe6, which is a decagonal (d) quasicrystal. The behaviours of k are typical for complex metallic alloys, i.e., a relatively small magnitude, a change of slope at about 50 K and an increase of the conductivity above 100 K. At room temperature the magnitude of k for all the samples is between 2.7 W/mK and 3.3 W/mK, which is comparable to that of thermally insulating amorphous SiO2 and Zr/YO2 ceramics. The reason for such a low thermal conductivity is because both, the electronic and lattice conductivity are low. The electronic contribution to the thermal conductivity is low because of the large electrical resistivity of the samples. The lattice thermal conductivity is greatly reduced because of the enhanced umklapp process of the phonon scattering (caused by the large lattice constant) and by the disorder in the structure. Keywords: complex metallic alloys, thermal conductivity, spectral conductivity Toplotna prevodnost (k) kompleksnih kovinskih zlitin Al73Mn27-iFex (x = 0, 2, 4, 6) je bila izmerjena v razponu temperatur od 2 K do 300 K. Vse zlitine, z izjemo Al73Mn21Fe6, ki je dekagonalni (d) kvazikristal, so Taylorjeve (T) faze. Vedenje toplotne prevodnosti k je značilno za kompleksne kovinske zlitine: relativno majhna velikost, sprememba naklona pri približno 50 K in povečana prevodnost nad 100 K. Pri sobni temperaturi je velikost k za vse preizkušance med 2,7 in 3,3 W/mK, kar je primerljivo s toplotno izolirno amorfno SiO2- in Zr/YO2-keramiko. Vzrok za tako majhno toplotno prevodnost sta majhni elektronska in mrežna prevodnost. Elektronski prispevek k toplotni prevodnosti je majhen zaradi velike električne upornosti zlitin. Mrežna toplotna prevodnost je zmanjšana zaradi povečanja procesa "umklapp" razpršitve fononov (zaradi velikega parametra mreže) in nereda v strukturi. Ključne besede: kompleksne kovinske zlitine, toplotna prevodnost, spektralna prevodnost 1 INTRODUCTION of the samples were annealed in argon at 900 °C and 930 °C for up to 698 h and subsequently quenched into The Al-Mn-Fe system contains several complex water2. All the samples were T (Taylor) phases of the metallic alloy phases, which have recently attracted composition Al73Mn27-xFex (x = 0,2,4,6), except for the increasing interest. Among them is the orthorhombic d-Al73Mn21Fe6, which was a decagonal quasicrystal. Taylor (T) phase, the structure of which is built of atomic The thermal conductivity k of the Al73Mn27-xFex (x = layers stacked along the [010]-direction. Along this axis, 0, 2, 4, 6) complex metallic alloys was measured in the pentagonal columnar clusters are formed For this temperature interval from 2 K to 300 K using an absolute reason, they are considered to be approximants of the steady-state heat-flow method. The thermal flux through decagonal (d) Al-Mn phases. The unit cell of the T-phase the samples was generated by a 1-kfi RuO2 chip-resistor, contains 156 atoms, with many of the sites having either glued to one end of the sample, while the other end was a fractional occupation or a mixed Al/Mn occupation, so attached to a copper heat sink. The temperature gradient that a great inherent chemical disorder exists on the across the sample was monitored by a chromel-gold lattice2. As part of a systematic investigation of the differential thermocouple (gold with 0.07 % of Fe)3. The transport properties of T-Al-Mn-Fe, here we present the electrical resistivity p (conductivity o =1/p) was results of the thermal conductivity measurements of measured between 300 K and 2 K using the standard T-Al73Mn27-xFex (x = 0,2,4) complex metallic alloys and four-terminal technique. for a comparison of the d-Al73Mn21Fe6 quasicrystals. 3 RESULTS AND DISCUSSION 2 EXPERIMENTAL The temperature dependence of the thermal conduc- The polycrystalline samples were produced from the tivity k(T) for all the investigated samples is shown in constituent elements by levitation induction melting in a Figure 1. The thermal conductivity of the T-phases and water-cooled copper crucible in argon atmosphere. Parts d-Al73Mn21Fe6 shows characteristic behaviour for the 6 S T-Al„Mn,T T-A!„Mn,,Fe, v T-Al„Mn„Fe^ W , r d-Al„Mii,,Fe, r r/K Figure 1: Temperature-dependent thermal conductivity k(T) between 2 K and 300 K for samples (T,d)-Al73Mn27-xFex (x = 0, 2, 4, 6) Slika 1: Odvisnost med temperaturo in toplotno prevodnostjo k(T) med 2 K in 300 K za zlitine (T,d)-Al73Mn27-xFex (x = 0, 2, 4, 6) complex metallic alloys (Mg32(Al,Zn)49, Al74Pd22Mn4, /3-Al3Mg2 and £-phases (Al-Pd-transition metal)4, which is a relatively small value, a change of slope at about 50 K and a rise of the conductivity above 100 K. At room temperature (300 K) the magnitude of the thermal conductivity is in the interval 2.7-3.3 W/mK for all the samples. Such a small magnitude of the thermal conductivity is a characteristic of thermal insulators like SiO25 and Zr/YO2 6. The small value of the thermal conductivity has also been found in the icosahedral quasicrystals j-Al-Pd-Mn 7,8, which is explained by the small value of electronic density of states at Ef (the small contribution of the electrons to the thermal conductivity) and no periodicity of the sample lattice (the small contribution of the phonons to the thermal conductivity). Our samples (T,d)-Al73Mn27-xFex (x = 0, 2, 4, 6) exhibit a very high electrical resistivity (Table 1) compared to simple metals (order of magnitude 1 ^fi cm), so the contribution of the electrons to the thermal conductivity is much smaller than the lattice (phonon) contribution. The low thermal conductivity can be qualitatively explained by the impact of the structure (disorder) of the studied samples on the thermal transport. Table 1: Values of the electrical resistivity p and the thermal conductivity K at room temperature Tabela 1: Električna upornost p in toplotna prevodnost K pri sobni temperaturi Samples p (^Q cm) K (W/mK) T-Al73Mn27 5071 2.69 T-Al73Mn25Fe2 2529 3.01 T-Al73Mn23Fe4 2283 2.83 d-Al73Mn21Fe6 720 3.28 has been described in detail in the previous investigation of the j-Al-Pd-Mn system9. The electrons (Ke) and lattice/phonon (ki) both contribute to the thermal conductivity K(r): k(t) = Ke(r) + Kl(r) (1) It is common practice to estimate the electronic part using the Wiedemann-Franz law Ke = L,T/p (2) where Lq = 2.44 x 10-8 W fi K-2 is the Lorenz number, and p is the electrical resistivity. Here we applied a more elaborate analysis based on the Kubo-Greenwood response theory i0,11,12. The central quantity of this formalism is the spectral conductivity function that incorporates both the band structure and the transport properties of the system. All the electronic contributions to the transport coefficients, including the resistivity, the thermopower and the thermal conductivity, are related to the spectral conductivity function. We have analyzed both the electrical conductivity and the thermopower to obtain the properties and the shape of the spectral conductivity function in the vicinity of the Fermi level. Our final results on the spectral conductivity function are shown in Figure 2. The details of the analysis are given in reference13. We should mention that this is a modified version of the procedure originally developed by Landauro and Macia 14,15,16 and is adjusted to suite the experimental data in this class of compounds. The thermal conductivity model appropriate for complex metallic alloys with a large-scale periodicity of the lattice and a small-scale atomic clustering structure Figure 2: The spectral conductivity function for all the samples. The singularity around the Fermi energy (the energy scale is shifted so that Ef = 0) is clearly pronounced. The sharpness of the pseudogap is directly related to the convex behaviour of the electric conductivity a(T) at low temperatures. Slika 2: Funkcija spektralne prevodnosti za vse zlitine. Singularnost pri Fermijevi energiji (lestvica za energijo je premaknjena tako, da je Ef = 0) je jasno pokazana. Ostrina psevdoreže je neposredno povezana s konveksnim vedenjem električne prevodnosti a(T) pri nizki temperaturi. The most important feature of the spectral function in Figure 2 is the pronounced pseudo gap around the Fermi level. Within the energy range of ±0.1 eV the spectral function loses around 40 % of its spectral weight. Moreover, our analysis, based on transport measurements at low temperatures, reveals the fine structure of the pseudo gap, featuring the \E\1/2 singularity at the Fermi level. Once calculated the spectral conductivity function can be used to determine the electronic contribution, Ke, to the thermal conductivity. The results of the different contributions at room temperature (RT) are presented in Table 2. The lattice part is significantly larger than the electronic part. This is because the electrical resistivity of these samples is very high, so consequently the electrons also have a small contribution to the heat transport. Table 2: Thermal conductivity k, electronic contribution Ke calculated by the model of spectral conductivity, electronic contribution Ke0 calculated by the WF law, difference between the two former electronic contributions in (%) and ratio effective and normal Lorentz number ieff/io at T = 300 K Tabela 2: Toplotna prevodnost k, elektronski delež ke, izračunan po modelu spektralne prevodnosti, in elektronski delež ke0 izračunan po WF-zakonu, razlika med obema v odstotkih in razmerje med efektivnim in normalnim Lorentzovim številom Leff/Lo pri T = 300 K. e d/t Sample K/ (W/mK) Ke/ (W/mK) Ke,o/ (W/mK) (Ke-Ke,o) /K (%) Leff/Lo T-Al73Mn27 2.69 0.18 0.15 1 1.25 T-Al73Mn25Fe2 3.01 0.43 0.29 4 1.40 T-Al73Mn23Fe4 2.83 0.41 0.32 3 1.30 d-Al73Mn21Fe6 3.28 1.30 1.02 9 1.30 K H = K H exp k B T (3) where K 0H is a constant. The Debye thermal conductivity is written as: K D = C D T ^ r(x) (ex -1)' dx (4) where Cd = kh3, v is the average velocity of sound, Od is the Debye temperature, t is the phonon relaxation time, x = hm/kBT, and hm is the phonon energy. The different phonon-scattering processes are incorporated into the relaxation time t(x) and we assume that Matthiessen's rule is valid, r"^ ^, where r"^ is a scattering rate related to the j-th scattering channel. In an analogy with the £-phases in Al-Pd-Mn9, we consider two dominant scattering processes in the investigated temperature interval (from 2 to 300 K). First, the scattering of phonons on the structural defects of the stacking-fault type with the scattering rate 7 a' 2 2 rj' = ^^^ ^^s (5) where a is a lattice parameter, y is the Grüneisen parameter and Ns is the linear density of the stacking faults. Although the WF law is not valid, the difference between the electronic contribution obtained by the spectral conductivity model Ke and the one obtain using the WF law Ke,o is only a few percent (except for d-Al73Mn2iFe6, where it reaches up to 10 %). So, we can still use the WF law to predict the electronic contribution to the heat transport as a rough approximation. The lattice contribution Ki = k - Ke is analyzed by considering the propagation of long-wavelength phonons within the Debye model and the hopping of the localized vibrations. This picture assumes that large atomic clusters of icosahedral symmetry strongly suppress the propagation of phonons in the lattice of complex metallic alloys. The exceptions are the long-wavelength acoustic phonons, for which this material is an elastic continuum, and fracton-like localized vibrations within the cluster substructure that can participate in the heat transfer via thermally activated hopping. In the simplest model, the hopping of localized vibrations is described by the single activation energy Ea, yielding a contribution to the thermal conductivity9 Figure 3: Temperature dependent lattice thermal conductivity Kl (filled symbols) together with the two contributions: the Debye contribution KD (dashed line) and the hopping contribution KH (dash-dot line) of T-Al73Mn27 and d-Al73Mn21Fe6 shown in different scales. Slika 3: Temperaturna odvisnosti mrežne toplotne prevodnosti Kl (polni znaki) z dvema deležema: Debyejev delež KD (črtkana črta) in delež "hopping" kh (črtopična črta) za T-Al73Mn27 in d-Al73Mn2iFe6, prikazani z različnima lestvicama The second scattering mechanism is the umklapp processes with the phenomenological form of the scattering rate pertinent to complex metallic alloys 9, ^üii = Bx" T4 and for the total scattering rate we get = r^f^ + ru^Jj. The Debye temperature of the investigated T-phases is not known; therefore, we have used the 9d value reported for the related icosahedral j-Al-Pd-Mn quasicrystals, were 6d was commonly found to be close to 500 K9,19. Since our k(T) data are available only up to 300 K, it turns out that the fit is insensitive to a slight change of this Od value, so a fixed Od = 500 K is used. The Debye constant Cd was also not taken as a free parameter, but was instead calculated using v = 4000 m s-1, a value determined for the j-Al-Pd-Mn from ultrasonic data. The results of a fitting procedure using the relation ki(T) = kd(T) + kh(T) to the experimental data are shown for T-Al73Mn27 and for d-Al73Mn2iFe6 in Figure 3. The parameters of the fitting procedure for all the investigated samples are shown in Table 3. From Figure 3 it is easy to see that the Debye contribution kd(T) has a maximum at about 50 K, although it becomes smaller at higher temperatures. A similar behavior is conventional for periodic structures, where the origin of such behaviour is in the phonon-phonon umklapp scattering pro-cesses21. Table 3: The fit parameters for the lattice thermal conductivity Kl = Kd+KH Tabela 3: Parametri ujemanja z mrežno toplotno prevodnost Kl = kd + KH Sample A (107 s-1 K-2) B (104 s) Kh0 (W/mK) Ea (meV) T-Al73Mn27 5.6 1.8 4.8 17.7 T-Al73Mn25Fe2 4.8 1.7 5.2 19.6 T-Al73Mn23Fe4 4.4 2.1 4.9 18.5 d-Al73Mn21Fe6 2.3 2.3 3.3 11.3 of 2 than the Ea of ^-Al-Pd-Mn9. This smaller Ea value reflects the considerably less steep k(T) increase at temperatures above 100 K for our samples compared to ^-Al-Pd-Mn. On the other hand, the above Ea values correlate with the inelastic neutron24 and x-ray25 scattering experiments on i-Al-Pd-Mn quasicrystals, where dispersionless vibrational states were identified for the energies higher than 12 meV. In quasicrystals such dispersionless states indicate localized vibrations and are considered to be a consequence of the dense distribution of energy gaps in the phonon excitation spectrum. This prevents extended phonons from propagating through the lattice, whereas localized vibrations may still be excited. Therefore, localized vibrations also appear to be present in the giant unit cell of (T,d)-Al73Mn27-xFex samples, where their origin may be attributed to the cluster substructure. 4 CONCLUSION We have investigated the thermal conductivity of (T,d)-Al73Mn27-xFex (x = 0, 2, 4, 6) samples, which show typical behaviour of k(T) for complex metallic alloys, i.e., relatively small values (thermal insulators), a change of the slope at about 50 K and an increase of the conductivity above 100 K. We have separated k(T) into the electron Ke(T) and lattice (phonon) Ki(T) parts, which both have small values. The electron part was determined by the spectral conductivity model, and it is much smaller than the lattice part. The reason is the very high electrical resistivity of all the samples, which reduces the electron thermal conductivity. The lattice's thermal conductivity is greatly reduced because of the enhanced umklapp processes of the phonon scattering (caused by a large lattice constant and, consequently, a small Brillouin zone) and by disorder in the structure. The parameter A, which describes the structural defects of the stacking-fault type, is close to 107 s-1 K-2 for all samples. It is possible to estimate from A the linear density of stacking faults Ns. If we take typical values for the lattice parameter a ~ 1.4 nm and the Grüneisen parameter y ~ 2, we obtain Ns = 10AvÄ2 / 7a2y2k,2= 0.8 pm1. This micrometer-scale Ns value is comparable to those reported for the ^-Al-Pd-Mn8 i-Al-Pd-Mn22 and the decagonal d-Al-Mn-Pd23. Therefore, the stacking-fault-like structural defects may be considered as the source of the phonon scattering at low temperatures in the (T,d)-Al73Mn27-xFex (x = 0, 2, 4, 6) samples. The parameters B and a define the phonon scattering by the umklapp processes in a phenomeno-logical way. From the fitting procedure we get a = 1, so the frequency and temperature dependence of the umklapp term is r~ mT3. For all the samples, the hopping contribution Kh becomes significant above 100 K. The activation energy Ea for all the samples is between 10 meV and 20 meV and is smaller by a factor Acknowledgement This work was carried out within the activities of the 6*^ Framework EU Network of Excellence "Complex Metallic Alloys" (Contract No. NMP3-CT-2005-500140), and has been supported in part by the Ministry of Science, Education and Sports of Republic of Croatia through the Research Projects: 035-0352826-2848 "Thermal and charge transport in highly frustrated magnets and related materials"; 035-0352826-2847 "Modelling physical properties of materials with marked frustration or disorder" and 119-1191458-0512 "Low-dimensional strongly correlated conducting systems". 5 REFERENCES 1 Klein, H.; Boudard, M.; Audier, M.; deBoissieu, M.; Vincent, H.; Beraha, L.; Duneau, M.: Philos. Mag. Lett., 75 (1997), 197-208 2 Balanetskyy, S.; Meisterernst, G.; Heggen, M.; Feuerbacher, M. Intermetallics, 16 (2008), 71-87 A. Bilu{i}; A. Smontara; D. Stani}; J. Paji}, Strojarstvo (2009), in press 4 A. Smontara, A. Bilu{i}, @. Bihar, I. Smiljani}, Properties and Application of Complex Metallic Alloys in Book Series on Complex Metallic Alloys, Esther Belin-Ferre (ed.), London WC2H 9HE, ENGLAND: World Scientific Publishing (UK) Ltd., 2 (2009), 113-147 5 D. M. Zhu, Phys. Rev. B, 50 (1994), 6053-6056 6R. Mevrel, J. C. Laizet, A. Azzopardi, B. Leclercq, M. Poulain, O. Lavigne, D. Demange, J. Eur. Cer. Soc., 24 (2004), 3081-3089 7 A. Bilu{i}, @ Budrovi}, A. Smontara, J. Dolin{ek, P. C. Canfield, I. R. Fisher, J. Alloys Compd., 342 (2002), 413-415 8 A. Bilu{ic', A. Smontara, J. Dolin{ek, P.J. McGuiness, H. R. Ott, J. Alloys Compd., 432, (2007), 1-6 9 J. Dolin{ek, P. Jeglič, P. J. McGuiness, Z. JagliciC, A. Bilu{iC, @. Bihar, A. Smontara, C. V. Landauro, M. Feuerbacher, B. Grushko, K. Urban, Phys. Rev., B 72, (2005), 064208-11 10 R. Kubo, Journal of the Physical Society of Japan, 12 (1957), 570-586 11 D. A. Greenwood, Proceedings of the Physical Society, 71 (1958), 585-596 12 G. V. Chester and A. Thellung, Proceedings of the Physical Society, 77 (1961) 1005-1013 131. BatistiC; D. StaniC; E. Tuti{; A. Smontara, Croatica Chemica Acta, 83 (2010), in press 14 C. V. Landauro Saenz, Ph.D. thesis, Fakultäfür Naturwissenschaften der Technischen Universität Chemnitz genehmigte (2002) 15 C. V. Landauro, E. Macia, and H. Solbrig, Phys. Rev. B, 67 (2003), 184206-7 16E. Macia, T. Takeuchi, and T. Otagiri, Phys. Rev. B, 72 (2005), 174208-8 17 R Escudero, J. C. Lasjaunias, Y. Calvayrac and M. Boudard, J. Phys.: Condens. Matter, 11 (1999), 383-404 18 @. Bihar, A. Bilu{iC, J. Lukatela, A. Smontara, P. Jeglič, P. J. McGuiness, J. Dolin{ek, Z. JagličiC, J. Janovec, V. Demange and J. M. Dubois, J. Alloys Compd., 407 (2006), 65-73 19 C. Wälti, E. Felder, M.A. Chernikov, H.R. Ott, M. de Boissieu and C. Janot, Phys. Rev. B, 57 (1998), 10504-10511 20 Y. Amazit, M. de Boissieu and A. Zarembowitch, Europhys. Lett., 20 (1992), 703-706 21 N. W. Aschroft i N. D. Mermin, Solid State Physics, Saunders College, Philadelphia, 1976 22 S. Legault, B. Ellman, J. Strom-Olsen, L. Taillefer, T. Lograsso, D. Delaney, Quasicrystals, Proceedings of the 6th International Conference, Tokyo, Japan, 1997 (world Scientific, Singapore, 1998), p. 475 23 M Matsukawa, M Yoshizawa, K. Noto, Y. Yokoyama, A. Inoue, Physica B, 263-264 (1999) 146-148 24 C. Janot, A. Magerl, B. Frick, M. de Boissieu, Phys. Rev. Lett., 71 (1993), 871-874 25 M. Kirsch, R. A. Brand, M. Chernikov, H. R. Ott, Phys. Rev B, 65 (2002), 134201-8