You are on page 1of 5

Materials Science and Engineering A271 (1999) 286 290 www.elsevier.

com/locate/msea

Lattice parameters of TM(3d)Al solid solutions


G. Shao *, P. Tsakiropoulos
School of Mechanical and Materials Engineering, Uni6ersity of Surrey, Guildford, Surrey GU2 5XH, UK Received 15 March 1999; received in revised form 4 May 1999

Abstract The role of valence electron density on the deviation of the apparent lattice parameter of Al in disordered TM(3d)Al solid solutions from the true value has been studied. It is shown that this deviation is proportional to the difference in the number of valence electrons between Al and the solvent metals and the trend breaks at the Cu Al solid solution. This is attributed to the complete occupancy of the d-band in Cu. 1999 Elsevier Science S.A. All rights reserved.
Keywords: Lattice parameters; Vegards law; Atomic diameter; Valence electrons

1. Introduction Four factors control the structures of alloys [1,2]. The rst three are the Hume Rothery rules, namely the size factor, the relati6e 6alency ( or electron concentra tion) factor and the electrochemical factor (electronega ti6ity). The fourth factor is the angular character of valence electrons, which affects the density of states in metals and the nature and angular disposition of hybrid bonds in compounds. The 6alence electrons are dened in modern terms as electrons whose energies are above the core level. Hume Rothery states that unless the solute and solvent radii lie within about 15% of each other, extensive solid solution cannot be formed even though other factors are favourable. This has been explained by calculating the elastic energy due to lattice distortion [3]. However, if the solute and solvent radii lie within 15% of each other, extended solid solution may or may not be formed, as determined by other controlling factors. Vegards law is a statement of the effect of the size factor on lattice spacing of a terminal solid solution phase. The law is only a rough approximation, as it is evident that the above factors strongly inuence the lattice spacings of solid solutions. Therefore, deviation of lattice parameters from Vegards law exists in most
* Corresponding author. Tel.: + 44-1483-300-800; fax: + 44-1483259-508. E -mail address: g.shao@surrey.ac.uk (G. Shao)

alloys. Fig. 1 shows experimental data [48] on the lattice parameters of some binary Ag alloys. Firstly, it is noticed that for a terminal solid solution phase of limited solubility, there exists approximate linear dependence of lattice parameter on the solute content. Secondly, even in systems such as AgAu, where a continuous solid solution phase is formed between Ag and Au, there is deviation of the lattice parameter from Vegards law. This demonstrates that the apparent atomic diameters (AAD ) of non-stable structures of pure elements, which are frequently derived by extrapolation from data of the stable terminal solid solution phases, seldom reect the true atomic diameters (AD ) of these non-stable elemental structures (see the deviation of the apparent lattice parameter of Au, aapp. from aAu in Fig. 1). The difference between AAD and AD is a measure of the deviation of lattice parameters of a solid solution phase from Vegards law. We need to determine the AD values of an element in different solvent structures in order to understand alloying behaviour. Therefore, it is necessary to eliminate the contributions of other factors such as the valency and electrochemical factors. Raynor [9] studied the valency factor of solid solutions of Cu, Ag and Au, and showed that the negative deviations of AAD from AD due to the valency factor in the solid solutions of the group IB elements (Cu, Ag and Au) are proportional to the square of the valency difference between the solvent and the solute (a group IB element) [9]. However, Raynors observations were based only on solute ele-

0921-5093/99/$ - see front matter 1999 Elsevier Science S.A. All rights reserved. PII: S 0 9 2 1 - 5 0 9 3 ( 9 9 ) 0 0 2 2 3 - 3

G. Shao, P. Tsakiropoulos / Materials Science and Engineering A271 (1999) 386290

287

about 3, 4 and 12% smaller than for atoms with z = 12 [3]. The co-ordination numbers in solids with metallic bonds are either 12 or 8 (see Table 3.2 in Ref. [3]). It will be shown below that for metals, the dependence of AD on z suggests that metal atoms tend to maintain their atomic volume in different structures. The average space each atom takes, Va, is: Va = W (zN0) 1 = n 1Vc = n 1a 3 c (1) where W is the molar atomic weight, z is the density, N0 is Avogadros number, Vc is the unit cell volume, n is the number of atoms in a unit cell, and ac is a characteristic crystallographic distance, which is equivalent to the lattice parameter of a cubic structure (i.e. the cube root of the unit cell volume). Thus, by changing z from 12 to 8, we get:
Fig. 1. Lattice parameters for AgX alloys. Notice the deviation of the apparent lattice parameter of Au, which is obtained by extrapolation from the Ag end, from the real value. Data from: Ag Al [4]; Ag Au [5]; Ag Cu [6]; AgMg [7] and AgZn [8].
(12) (12) 1/3 1/3 a (8) ] fz f c = [V a n rp

(2)

ments, which are on the right of the group IB in the periodic table, where valencies of elements were well dened. Thus, it is difcult to make use of the results for solid solutions in transition metal elements where there is great difculty in dening the values of valencies in the conventional way. In this work, we study the effects of Al on the lattice parameters of disordered solid solutions in TM(3d) Al alloys.

2. Lattice parameters of solute element of a non-stable structure The rst task is to dene the AD values of an element in different structures, as the AD of a metallic atom depends on the co-ordination number z. It is well known that the radii of atoms with z = 8, 6 and 4 are
Table 1 Calculated and experimental [10] ac Element ac (101 nm) Calculated (RTa) Fe (fcc)c Fe (bcc)b Al (fcc)b Al (bcc)c Ti (bcc)c Zr (bcc)c Cr (fcc)c
a

where fz denotes the factor due to the change in coordination number and frp denotes the packing fraction relative to that with z = 12. We have fz : 0.97 and 1/3 1/3 f = (0.734/0.68)1/3 : 1.03, giving fzf : 1, for rp rp structural change between z = 12 and z = 8. Small volume change was experimentally observed for the transformation between fcc and bcc Fe ( 0.08%), suggesting that a reasonably good approximation of AD can be achieved using Eq. (2). Table 1 compares the ac values, which are calculated by Eq. (2) with experimental ones. It is noticed that the calculated values in Table 1 agree very well with the experimental values. The differences between the calculated room temperature (RT) values and experimental values above RT (fcc Fe and Cr as well as bcc Zr and Ti) are negligible, when considering a thermal expansion coefcient of 10 5 1 K , which is typical for transition metals (Table 1). It is worth mentioning that Eq. (2) works better than the Paulings AD values (reviewed in Ref. [3]) for elements whose room temperature structure is A3 (e.g. Ti, Zr). This is largely due to the fact that most hexagonal

Linear expansion coefcient (106 K1) Experimental 3.639 (916C) 2.866 (RT) 4.045 (RT) 3.306 (900C) 3.27 (RT) 3.609 (862C) 3.68 (\1850C) 11.7 23.6 9.7

3.61 3.21 3.282 3.597 3.65

Room temperature. Room temperature stable structures. c Structures which are not attainable at room temperature via equilibrium processing routes.
b

288

G. Shao, P. Tsakiropoulos / Materials Science and Engineering A271 (1999) 286290

Fig. 2. Lattice parameters for some TM(3d)Al alloys. The data for Ti Al are the ac value derived from the lattice parameters of the stable hexagonal solid solution phases (A3 and DO19 structures). Data from: Ti Al [12]; VAl [13]; CrAl [14] and FeAl [15,16].

available in the literature [1216]. The ac value for TiAl is derived from the lattice parameters of the stable hexagonal solid solution phases (A3 and DO19 structures). As discussed in the previous section, the ac values of TiAl can be considered as the lattice parameters of the equivalent bcc structure. It is noticed that when Al contents exceed critical values in both the TiAl and FeAl systems, the lattice parameters deviate drastically from straight lines, which are followed by low solute content alloys. By consulting the phase diagrams, it is clear that these critical Al contents correspond to the boundaries beyond which the solid solution phases in both systems undergo chemical ordering to form compound structures. Therefore, the drastic deviations from the straight lines in the cases of TiAl and FeAl are due to the introduction of covalent and/or ionic bonds in the solid phases. Fig. 3 shows a plot of the AD values of metals for both covalent and metallic bonds, using data from Table 3.2 of Ref. [3]. It suggests that for the same element, AD (covalent) is smaller than AD (metallic). A linear correlation between AD (covalent) and AD (metallic) in Fig. 3 can be described by: ,) AD (covalent) = 0.086 + 0.831AD (metallic) (A (3)

structures are not ideal, leading to a split of the ideal 12 nearest neighbours into six nearest and six second nearest neighbours, making it difcult to dene AD in Paulings case. For example, Paulings AD (z = 12) , , this gives ac = 3.248 A , for bcc value for a-Ti is 1.45 A Ti, taking into account the coordination number change. In order to reconcile the experimentally determined lattice parameter of bcc Ti at 900C, we need a linear expansion coefcient of 64 10 6 K 1, which is too large for a transition metal of a close packed structure. On the other hand, using Eq. (2), we can ,, calculate the lattice parameter for bcc Ti as ac = 3.28 A which is much closer to the experimental value at 900C (Table 1). Extrapolation of experimental data for Ti AlV alloys gives an apparent lattice parameter for bcc , , which is quite close Ti at room temperature as 3.27 A to the calculated true value using Eq. (2) [11]. In summary, metals tend to conserve their atomic volume, and hence tend to maintain constant densities in different crystal structures. It is worth pointing out that in a disordered metallic terminal solid solution, little change in the nature of bonding is anticipated, as an evident change in the nature of bonding would introduce chemical ordering in the solid solution phase.

The AD value for an ionic metal is smaller than AD (covalent). Therefore, ordering due to the formation of either covalent or ionic bonding in a solid solution phase will lead to a further decrease in lattice parameters. The effect of the valency factor on the lattice parameters can be studied by using only the low solute content parts of ac vs at% Al plots (where the solid solution phases are disordered and hence the metallic bond is dominant) to extrapolate for the apparent lattice parameters of Al in different solvents. It is shown in Fig. 2 that the apparent lattice parameters of bcc Al derived from different alloys are also different. Also, all the apparent bcc lattice parameters of Al shown in Fig. , (Table 1). 2 are smaller than the true value of 3.21 A

3. Lattice parameters of TM(3d) Al alloys The true lattice parameters of bcc and fcc Al are given in Table 1. We are now ready to study the effect of Al content on the lattice parameters of some TM(3d)Al alloys. Fig. 2 shows the lattice parameters of some bcc alloys, using experimental data which are

Fig. 3. AD of the covalent bonds plotted against AD of metallic bond. Data are from Table 3.2 of Ref. [3].

G. Shao, P. Tsakiropoulos / Materials Science and Engineering A271 (1999) 386290

289

Fig. 4. l vs De /a. Sources of data are: TiAl [12]; VAl [13]; Cr Al [14]; Fe Al [15,16]; CoAl [17]; NiAl [18] and CuAl [19].

We dene the percentage deviation of the apparent lattice parameters from the true lattice parameters of Al of different structures as l = 100(a aapp)/a, where a is the true lattice parameter. Using available experimental data [1219], we can plot l against the difference in electron per atom ratio, De /a ( = e /a (TM) e /a (Al)) (Fig. 4). The De /a value reects the efciency of Al addition to the change of e /a in a particular TMAl alloy. Thus the l value is a measure of the maximum deviation of the lattice parameters of the disordered solution phase from Vegards law for each TMAl alloy system. For all of the TM(3d) Al alloys but CuAl, the l values of Al are almost in perfect linear relationship with De /a. It should be emphasised that the relationship in Fig. 4 is not affected by the accuracy of the estimated AD value for bcc Al, since the latter is only used as a reference. The l value of Al for the Cu Al solid solution deviates drastically from the linear l De /a relationship. Actually, the l values of Al in all the group IB transition metal solid solutions are small. The l values of Al are 2.2% [4] and 2.38% [20] in Ag and Au solid solutions, respectively. This suggests that for group IB transition metals, it is incorrect to use the total number of non-core electrons to account for the effect of the valency factor on l. An important difference between group IB elements and other TM is that the group IB elements have a completely lled d-band. As the d-electrons conned in completely occupied d-band have lower energy levels than the electrons in the partially lled s-band, their mobility is drastically lower (higher energies are needed to excite the d-electrons into the s-band and the d-electrons are then less free) than the d-electrons of the TM elements on the left of group IB. Actually, the d-band occupancy makes the group IB elements different in many ways from other TM elements [3]. For all other TM elements on the left of group IB, their d-bands are only partially occupied so that a very small energy input ( kT ) can pump a

d-electron up to a higher energy state in either the d-band or the s-band (d- and s-bands overlap). Therefore, it is not surprising that Al in CuAl has remarkably lower l value than in NiAl, owing to the low effective number of mobile electrons per atom, i.e. e /a = 1 (the valency of Cu dened in the conventional way), for Cu. First principle calculations of the selfconsistent valence s and d energy levels across the 3d and 4d TM series have shown that the energies of s and especially d states decrease rapidly once the d shell is lled in the group IB elements (see Fig. 5 of Ref. [21]). In this work, we have focused on the solvent-end disordered solid solutions; the TM(d)Al(p) hybridisation and/or the electrochemical factor are not considered to be the principal causes for the dependence of l values of Al in TM(3d)Al alloys on De /a. Theoretical calculations of charge distributions in the bcc structure of different TM(3d)Al alloys using, for example, the full potential linear mufn-tin orbital (FP-LMTO) method could help to provide a better understanding of the experimental results. As to the role of electronegativity, there is no trend in a plot of l against the electronegativity differences (Dx ) between TM(3d) and Al. In fact, the largest Dx value for Al in Cu corresponds to a considerably smaller l value than that, for example, in Ni. The dependence of l on De /a may be attributed to the tendency for Al to retain its original spatial charge density. Eq. (2) suggests that the atomic volume of a metallic element hardly changes due to structural variation by changing the co-ordination number between 12 and 8. When an Al atom is embedded in a TM (such as V and Nb) lattice, which has a smaller AD than Al, the charge density of the Al atom will be increased due to the reduction of its atomic volume. To sustain its original charge density, the Al atom could give some of its electrons to the surrounding TM atoms, leading to a smaller AAD. Experimentally measured charge transfer values in VAl and NbAl alloys are consistent with this, as Al has been found to give charge to V and Nb and charge transfer in NbAl is higher than in VAl [22]. This is also supported by the higher l value of Al in Cu than in Ag or Au, as the lattice parameters of the latter group IB elements are much closer to that of Al (in fact, the lattice parameters of Ag and Au were about a 7 thousandth larger than that of Al). Also, when Al is added to a TM element such as Ti, which has a larger AD than Al, the above argument would suggest that Al should gain charge from the surrounding TM atoms. The higher electronegativities of the TM(3d) elements, however, exert a barrier for charge transfer in this direction. The overall effect is that the l values of Al in such TM elements would be smaller. Further work on charge transfer measurement using the method proposed by Diplas and co-workers [22,23] would be useful in clarifying the points presented here.

290

G. Shao, P. Tsakiropoulos / Materials Science and Engineering A271 (1999) 286290 [2] A.P. Sutton, Electronic Structure of Materials, Oxford Science, Oxford, 1993. [3] A. Cottrell, An Introduction to Metallurgy, 2nd edn., The Institute of Materials, 1975. [4] F. Foote, E.R. Jette, Met. Tech. 7 (1940) 1229. [5] G. Sach, J. Weerts, Z. Phys. 60 (1930) 281. [6] N. Ageev, G. Sach, Z. Phys. 63 (1930) 293. [7] N. Ageev, V.G. Kuznetsov, Izv. Akad. Nauk SSSR (Khim) (1937) 289. [8] H. Lipson, N.J. Petch, D. Stockdale, J. Inst. Met. 67 (1941) 79. [9] G.V. Raynor, Trans. Faraday Soc. 45 (1949) 698. [10] W.B. Pearson, Handbook of Lattice Spacings and Structure of Metals, vol. 1, Pergamon, Oxford, 1958. [11] G. Shao, A.P. Miodownik, P. Tsakiropoulos, Phil. Mag. A 71 (1995) 1389. [12] E.S. Bumps, H.D. Kessler, M. Hansen, J. Metals 4 (1952) 609. [13] O.N. Carlson, D.J. Kenney, H.A. Wilhelm, Trans. Am. Soc. Met. 47 (1955) 520. [14] A.J. Bradley, S.S. Lu, Z. Krist. A96 (1937) 20. [15] A.J. Bradley, A.H. Jay, J. Iron Steel Inst. 125 (1932) 339. [16] A.J. Bradley, A.H. Jay, Proc. R. Soc. A136 (1932) 210. [17] H.L. Luo, P. Duwez, Can. J. Phys. 41 (1963) 758. [18] A.J. Bradley, A. Taylor, Proc. R. Soc. A159 (1937) 56. [19] A.J. Bradley, H.J. Goldsmith, J. Inst. Met. 65 (1939) 389. [20] E.A. Owen, E.W. Roberts, J. Inst. Met. 71 (1945) 213. [21] D.G. Pettifor, in: R.W. Cahn, P. Haasen (Eds.), Physical Metallurgy, Elsevier, Amsterdam, 1983. [22] S. Diplas, G. Shao, S.A. Morton, P. Tsakiropoulos, J.F. Watts, Intermetallics 7 (1999) 937. [23] S. Diplas, P. Tsakiropoulos, R.M.D. Brydson, J.F. Watts, Phil. Mag. A 77 (1998) 1067.

4. Conclusions This work examines the dependence of l, the percentage deviation of the apparent lattice parameters of solute Al in TM(3d) Al solid solutions from the true ones, on valence electron densities. As long as the d-band is not completely lled, l is found to be proportional to De /a = e /a (TM) e /a (Al). The lower l value of Al in Cu is attributed to the low effective number of mobile electrons in Cu, as its d-electrons are conned in a lled d-band. A method has been proposed to calculate the lattice parameters of metallic solute elements of a non-stable structure. These, together with l, can be used to realistically estimate the lattice parameters of disordered TMAl solid solutions. It is suggested that a drastic decrease of lattice parameters from the linear dependence of lattice parameters on solute content is evidence of chemical ordering in the solid phase.

References
[1] A. Cottrell, Introduction to the Modern Theory of Metals, The Institute of Metals, London, 1988.

You might also like