Thermodynamic Properties of the Cu-A1 System Correlation with Bonding Mechanisms BY JOHN HAIRAND D. €3. DOWNIE" Dept. of Metallurgy University of Strathclyde Glasgow G1 1XN Received 11th October 1973 A wide-ranging study of the thermodynamic properties of the Cu-A1 system at 773 K has been carried out. An e.m.f. technique using solid CaF as electrolyte was used to obtain the partial free energies of Al in a series of selected alloys. Enthalpies of formation (A&) were measured using a liquid-tin solution calorimeter and combined with integral free energies (AGf) to evaluate the excess entropies(ASP). Volume changes on alloying (A Vf) were calculated from existing room temperature lattice parameters. Plots of AHf AS? and A Vf against composition show parallel relationships particularly between ASfs and A Vf.Due to lack of heat capacity data for the alloys it is not possible to separate vibrational and configurational components of entropy but reasons are given for assuming that both may be related to volume changes. Assuming volume contraction is associated with orbital overlap it is concluded that covalency is present in all the intermediate phases to varying extents. There is electronic spectroscopic evidence to support the view that there is an ionic contribution to bonding in the a-phase alloys which diminishes with increasing Al-content beyond this range. From other electron spectroscopic data it is deduced that the covalent bonding in the yz-phase is largely due to orbital overlap while that of the t2-and 0-phases arises mainly from charge localisation.In recent years some attention has been given to the thermodynamic behaviour of intermediate phases in binary metallic systems in relation to the bonding between the specie^.^'^ The Cu-A1 system is one which is characterised by the formation of a large number of intermediate phases some of which are stable only within fairly narrow temperature ranges. However at temperatures around 773 R,at which thermodynamic parameters can conveniently be measured the presence of five inter- mediate and two terminal phases is reported by Hansen and Anderko.6 Since hitherto the Cu-A1 system had not been examined in detail it was decided to make a comprehensive thermodynamic study of the system involving free energy and enthalpy of formation measurements to provide a basis for intercomparison of phases and their possible bonding mechanisms.EXPERIMENTAL ALLOY PREPARATION Alloys of selected compositions were preared by melting under argon in a high-frequency furnace the appropriate weights of the component metals contained in a silicon carbide crucible. The melts were solidified and remelted twice prior to removal from the furnace to ensure a reasonable degree of homogenization. The solid alloys were then wrapped in copper foil sealed in evacuated silica capsules and annealed at 773 K for periods of up to three weeks to effect a greater degree of homogenization and also to ensure the presence of the equilibrium phases. FREE ENERGY MEASUREMENTS The partial free energies of mixing of the Al-component in the alloys were obtained using the solid-electrolyte galvanic cell Al AIF,/CaF2/AIF, AI(Cu) 56 J.HAIR AND D. B. DOWNIE and the relationship Ac~l=-LYE where AGA is the parameter required 2is the state of ionisation (assumed equal to three) F is the Faraday constant (96 487 C/g-atom) and E is the measured e.m.f. Since electrode materials for the galvanic cell were required in an extremely fine particle size the alloys were pulverised in a percussion mortar and the resulting powder intermixed with pre-dried aluminium fluoride using an agate mortar and pestle. The electrodes were formed by pelletizing the powder in a steel die to form discs of about 8 mm dim. and 3 mm thickness and subsequently sintering under argon at 773 K.The reference electrodes were similarly produced from high purity X-ray grade A1 powder. After assembly argon gas was allowed to pass around the cell for one hour before the furnace was switched on. For each alloy sufficient time (ranging from 24 to 48 h) was allowed for the reaction to come to equilibrium before readings were taken. High initial temperatures were chosen in order to encourage equilibrium conditions. In each case the cell was cycled within a 100-150"range in 50" intervals until results of acceptable constancy at each temperature were achieved. ENTHALPY MEASUREMENTS The isoperibol calorimeter used to measure the enthalpies of formation of the alloys has been employed extensively in the Department of Metallurgy at Strathclyde University to make similar measurements on a number of other binary alloy Essentially it is liquid tin solution calorimeter based on the design of Orr Goldberg and H~ltgren.~ Standard procedures for operation were followed.In order to maximise solution rates the temperature of the liquid tin bath was set as high as possible i.e. in the region of 780 K. However with stoichiometric compounds in the composition range 38-50 at. % A1 solution rates were still rather low. Since the problem appeared to arise from a combination of low density and compound stability a number of pelletising techniques O designed to increase both the density and the reacting surfaces of the samples were tried. These attempts resulted in some slight increase in solution rates but to effect further improvement the calorimeter was modified to aIIow operation at higher bath temperatures.This was achieved by re- designing the copper jacket furnace and incorporating a solid state controller used in con- junction with a platinum resistance thermometer.10 This allowed bath temperatures of up to 950 K to be used resulting in greatly increased solution rates. The heats of solution of the pure metals in liquid tin were obtained first of all using lump samples of approximately 0.10g weight. In a similar fashion the heats of solution of the alloys were measured for a preheat temperature of 773 K. Using these results and existing heat capacity and enthalpy of fusion data,ll the enthaipies of formation of the selected alloys at the preheat temperature were obtained.RESULTS FREE ENERGIES Over the relatively narrow range of temperature involved in the measurements (viz. 700-850 K) it was assumed that a linear relationship holds between the measured e.m.f. and the temperature i.e. E = Eo+(dE/dT)T. For each alloy the values of Eo and dE/dT were calculated from the experimental data by the method of least squares. These values are reported in table 1 along with the derived values of AGAlfor a temperature of 773 K. The estimated average error in these values is approximately 6 %. The only published results of free energy measurements on alloys from the Cu-A1 system are those of Ali Samokhval Geiderikh and Vecher.12*l3 These results are plotted along with the present ones against composition in fig.1. Ali et al. used e.m,f. cells of two types a solid electrolyte cell in the temperature range 933-1033 K and a liquid electrolyte in the range 735-825 K. Their results in the a-phase region 58 THERMODYNAMIC PROPERTIES OF Cu-A1 SYSTEM TABLE PROPERTIES OF Cu-A1 ALLOYS AT 773 K 1.-THERMODYNAMIC -(WdTI/ AGAi/ Act/ AH,/ AScI AS I phase xA1 EObV mV K-1 kJ/g-atom kJ/g-atom kJ/g-atom kJ K-'/g-atom J K-t;g-atom composition 0.016 324.3 0.084 -112.67 -1.97 --0.070 237.0 0.075 -85.40 -7.20 -6.19 1.30 -0.79 0.140 229.2 0.039 -75.10 -12.67 -9.15 4.55 + 1.21 0.220 267.2 -0.074 -60.76 -13.92 -0.273 250.8 -0.049 -61.60 -18.43 -0.310 270.5 -0.081 -60.14 -22.34 -20.29 2.65 -2.50 0.340 245.0 -0.117 -44.75 -23.68 -21.69 2.57 -2.76 0.363 121.8 -0.039 -26.56 -24.10 -21.19 3.76 -1.68 0.380 99.8 -0.015 -25.51 --0.395 --20.79 * -24.14 -20.67 4.48 -1.08 -0.425 94.1 -0.050 -16.06 --0.445 --12.79 * -23.39 -20.40 3.87 -1.84 0.468 56.9 -0.03 1 -9.53 --0.490 --8.68 * -22.09 -19.92 2.81 -2.95 -0.520 40.4 -0.021 -6.98 -17.05 -0.615 42.5 -0.026 -6.52 --0.670 4.3 0.008 -3.01 -16.63 -13.05 4.66 -0.61 0.722 3.0 O.OO0 -0.88 -12.33 -0.840 2.7 O.OO0 -0.79 -6.53 -0.950 2.9 O.Oo0 -0.84 -1.62 -0.996 --0.75 -0.17 0.75 +0.53 * values obtained by interpolation.have been extrapolated from the higher temperature range to 773 K and referred to solid A1 as standard state using published data l1 to calculate the free energy of solidification of A1 at 773 K. The phase field limits shown in fig.1 are taken from Hansen and Anderko.6 -20 Ot -1201 1 I I I 0 0.2 0.4 0.6 0.8 1.0 XAl FIG.1.-Partial free energies of mixing of Al in CU(AGA) as a function of composition ; 0,present results ; x,Ali et af. J. HAIR AND D. B. DOWNIE Fig. 1 shows that there is good agreement between the two sets of results in the yz-and (rz+@-phase fields but elsewhere agreement is only fair to poor. In view of their reported lack of reproducibility in e.m.f. values and the uncertainty in the degree of ionisation when using the fused chloride cell in the lower temperature region and also of the large extrapolation of their high temperature measurements the present results are preferred throughout. Fig. 1 also shows that neither set of results includes alloys in the 6- c2-and q2-phase fields.In the case of the present authors the results from alloys within these phase fields were very erratic and considered unreliable. The reason for this is not clear but it may be associated with the difficulty in establishing the equilibrium phase when it is confined to very narrow compositional limits. As a result the AGA values for the compositions from these phase fields shown in table 1 have been ob- tained by interpolation. The integral free energies of formation ((AG,) which are also shown in table 1 were obtained from the partial values using the Cibbs-Duhem relationship. A I 0 0.2 0.4 0.6 0. 0 a XAl FIG.2.-Enthalpies of formation of Cu-Al alloys as a function of composition 0,present results (773 K) ; v,Oelesen and Middel (298 K) ; A,Sinvhal and Khangoankar (303 K) ; .,Kubas-chewski and Heymer (598 K).ENTHALPIES The enthalpy of formation values (AHf)shown in table I are the averages of several determinations at a single preheat temperature. The average error is estimated at 2 %. For purposes of comparison the enthalpy values from table 1 have been plotted along with the published results of other workers as a function of composition in fig.2. The values of the present workers are joined by straight lines to show the trend. From the plot it is clear that overall there is good agreement amongst workers. Agree-ment is particularly good between the present values and those of Oelsen and Middel l4 which were obtained by mixing the liquid metals and evaluating the heats of formation at 298 K.This agreement and the general. accord with the values of the other workers THERMODYNAMIC PROPERTIES OF Cu-A1 SYSTEM for temperatures of 303 K (Sinvhal and Khangaonkar 15) and 598 K (Kubaschewski and Heynier 16) indicate that ACp values are very small. Unpublished results of Wittig l7 measured at 745 K but not included in fig. 1 are considerably less exo- thermic in the composition range 20-45 at. % Al but are in good agreement with the other results outwith this range. ENTROPIES AND VOLUME CHANGES The integral excess entropies of mixing have been derived from the appropriate free energy and enthalpy values obtained in this investigation. The results are recorded in table 1 and plotted against composition in fig.3. Volume changes on alloying have been calculated from the room temperature lattice parameter and crystal structure data of Pearson18 and of Westrnan.lg They are expressed as percentage change in volume over the "ideal " value i.e. the weighted sum of the volumes of the pure metals. The values obtained are plotted in fig. 3 for comparison with the excess entropies of formation of the alloys. XAI FIG.3.-Excess entropies of formation (AS?) and volume changes on alloying (A Vf)as functions of composition 0, ASfs ; X A Vf. DISCUSSION THERMODYNAMIC PARAMETERS At a temperature of 773 K the stable phases in the Cu-A1 system according to Hansen and Anderko,6 are a 0-19.6 at. % Al terminal solid. solution y2 30.9-37.5 at.% Al based on Cu,Al, 6 38.8-39.9 at. % Al based on Cu3A12, r2 43.9-44.8 at. % Al based on Cu,Al, q2 48.7-50.0 at. % Al based on CuAl 6 67.0-67.6 at. % Al based on CuAl, K 98.4-100.0at. % Al terminal solid solution. J. HAIR AND D. B. DOWNIE It has been suggested that the y2-and b-phases may be a single one with varying lattice constants ; the results of this investigation neither support nor contradict this suggestion. If the AHf values of the phases are taken as a measure of their relative stabilities the plot in fig. 2 suggests that the y2-phase around a composition of 34 at. % is the most stable closely followed by the 6- c2-and q,-phases with the &phase showing a much less negative value. This trend is followed approximately by the AV values in fig.3 indicating some correlation between bond strength and volume con-traction. An even more marked parallelism in compositional trends is shown by AV and AS,""values (fig. 3) and the question arises as to whether the relationship is due to varying configurational or vibrational entropy factors. Since all of the intermediate phases have positive entropies of formation (see AS values in table 1) then either the phases are not completely ordered at 773 K or there are positive contributions from vibrational sources. In the absence of the relevant heat capacity data for the alloys it is not possible to evaluate the vibrational contributions to entropy but it seems unlikely that the negative volume changes obtained on alloying would be associated with increases in lattice vibrations.The dilemma could be resolved by postulating sufficiently large positive contributions from electronic sources. In this case the varying vibrational contributions alone would be responsible for the parallelism with the AV values. However since there is no evidence for this unusual electronic behaviour it is more likely that the intermediate phases are not completely ordered and that the parallelism though still mainly associated with changes in vibrational properties may also be related to varying degrees of order. BONDING MECHANISMS Negative heats of formations in compounds are usually associated with ionic (i.e. charge transfer) or covalent (i.e. charge sharing) bonding. Thus in the Cu-A1 system in which negative heats of formation obtain throughout one should look for electronic behaviour which is consistent with these types of bonds.Fuggle,20 using X-ray photoelectric spectroscopy (XPS) has found that a-alloys in the CuAl system are ionic in character this resulting from charge transfer from A1 p-states to available states in the Cu valence-band. These are probably d-states made available by hybridization of the Cu atoms as proposed by Engel,21 to give average outer shells of 3d8.54s1p1-5instead of 3d104s1. This hybridization of Cu is not in accord with the results of Baer et al. in their electron-spectroscopy studies of transition metals 22 in which they found the Cu d-levels to be completely filled. However if Engel's proposal is accepted it is possible to explain the relatively wide compositional stability range of the CI solid solution compared with the low solubility of Cu in Al as shown by the K terminal solution in terms of availability of d-states which will be much less in the Al-rich alloys due to the greater separation of the Cu-atoms preventing hybridization.Also the asymmetric shape of the (AH, composition) curve (fig. 2) would result from the increased ionic contribution to bonding in the Cu-rich alloys. Covalentbonding can arise from two different types of valence electron behaviour (i) orbital overlap and (ii) localization of charges between the atoms. The former is accompanied by volume contraction and the latter results in reduction of valence band width. The fact that all the alloys in the Cu-A1 system form with a reduction in volume would suggest that covalency due to orbital overlap is present in all phases.THERMODYNAMIC PROPERTIES OF Cu-A1 SYSTEM At present investigation of electronic structures has been carried out only on two inter- mediate Cu-A1 alloys viz. Cu9A14 &-phase) and CuA1 (&phase). Using XPS Fuggle et aZ.23have determined that although Cu9A14 shows no reduction in valence band width CuA1 shows considerable reduction. It might be expected therefore that the increased bonding effect due to the charge localization would result in CuAl having the more exothermic heat of formation whereas the reverse is true (fig. 2). However the volume contraction of Cu9A14 is considerably greater than that of CuA1 (fig.3) signifying a greater degree of orbital overlap and as postulated in the preceding paragraph CugA14 being richer in Cu is also likely to have a greater amount of electron transfer resulting in the more exothermic heat of formation. In the absence of information on the electronic behaviour of the other phases the contribution of localization of charges to bonding cannot be assessed. However the AV' values suggest that the orbital overlap of the 8-phase is similar to that of yz while c2 and q2 have much less. The anonialously high (i.e. small contraction) value of AVf for the [,-phase compared to its AHf value (fig. 2) may be due to the presence of considerable charge-localisation as in the @phase. The foregoing discussion has shown that whereas some correlation between thermodynamic and electronic behaviour of alloys in the Cu-A1 system is possible complete analysis requires further experimental data.Information on the electronic behaviour of the 8- c2-,q2-and Ic-phases is required while on the thermodynamic side the low temperature heat capacities of all the phases would allow more thorough examination of the entropies of formation. Investigations to elucidate the electronic behaviour of the neglected phases are presently being pursued by XPS methods in the Department of Metallurgy in the University of Strathclyde and it is hoped to commence work on the cryogenic calorimetric aspects in the same place in the near future. The authors acknowledge the informa tion and helpful discussion provided by their colleagues Dr.J. C. Fuggle and Dr. L. M. Watson particularly on the electronic aspects of the investigation. One of the authors (J. H). also thanks the Science Research Council for provision of a maintenance grant. 0. Kubaschewski and W. Slough A Tentative Analysis of the Bond Types in Binary Metallic Systems (N.P.L. London March 1972). V. A. Geiberikh Russ. J. Chem. 1971,45 1083. W. H. Skelton N. J. Magnani and J. F. Smith Met. Trans. 1971 2,473. A W. Bryant W. G. Bugden and J. N. Pratt Acta Met. 1970 18 101. G M. Lukashenka Russ.J. Phys. Chem. 1968,42,405. M. Hansen and K. Anderko Constitution of Binary Alloys (McGraw-Hill New York,2nd ed. 'G. 1958). R. Blair and D. B. Downie Metal Sci. J. 1970 4 1. * R. A. Come11 and D.B. Downie Metal Sci. J. 1973 7 12. R. L. Orr A. Goldberg and R. Hultgren Rev. Sci. Instr. 1957 28,767. lo John Hair Ph.D. Thesis (Univ. Strathclyde) to be submitted. R. Hultgren R. L. Orr and K. K. Kelley Supplement to Selected Values of Thermodynamic Properties of Metals and Alloys (Lawrence Radiation Laboratory Univ. California Berkley U.S.A.). l2 S. A. Ali V. V. Samakhval V. A. Geiderikh and A. A. Vecher Russ. J. Phys. Chem. 1972,46 139. l3 S. A. Ali V. V. Samakhval V. A. Geiderikh and A A. Vecher Rum. J. Phys. Chem. 1973,47 22. l4 W. Oelsen and W. Middel Mitt. KWI Eisenforsch. Diisseldorf 1937 19 1. l5 R. C. Sinvhal and P. R. Khangaonkar Trans. Indian Znst. Metals 1967 (June) T.P.421 107. l6 0. Kubaschewski and G. Heymer Trans.Faraday SOC., 1960 56,473. J. HAIR AND D. B. DOWNIE l7 F. E. Wittig Univ. Munich Germany (private communication). W. B. Pearson A Handbook of Lattice Spacings and Structures of Metals and Alloys (Pergamon, London 1958). l9 S. Westman Acta Chem. Scand. 1965 19 2371. '* J. C. Fuggle Dept. of Metallurgy Univ. Strathclyde (private communication). 21 N. Engel Acta Met. 1966 15 557. 22 Y.Baer P. F. Heden J. Hedman M. Klasson and K. Siegbahn Physica Scripta 1970,1,55. 23 J. C. Fuggle L. M. Watson D. J. Fabian and P. R. Norris Solid State Comm. 1973 13 507.