Re-investigation of phase equilibria in the system Al-Cu and structural analysis of the high-temperature phase η1-Al1-δCu.

The phase equilibria and reaction temperatures in the system Al-Cu were re-investigated by a combination of optical microscopy, powder X-ray diffraction (XRD) at ambient and elevated temperature, differential thermal analysis (DTA) and scanning electron microscopy (SEM). A full description of the phase diagram is given. The phase equilibria and invariant reactions in the Cu-poor part of the phase diagram could be confirmed. The Cu-rich part shows some differences in phase equilibria and invariant reactions compared to the known phase diagram. A two phase field was found between the high temperature phase η1 and the low temperature phase η2 thus indicating a first order transition. In the ζ1/ζ2 region of the phase diagram recent findings on the thermal stability could be widely confirmed. Contrary to previous results, the two phase field between δ and γ1 is very narrow. The results of the current work indicate the absence of the high temperature β0 phase as well as the absence of a two phase field between γ1 and γ0 suggesting a higher order transition between γ1 and γ0. The structure of γ0 (I-43m, Cu5Zn8-type) was confirmed by means of high-temperature XRD. Powder XRD was also used to determine the structure of the high temperature phase η1-Al1-δCu. The phase is orthorhombic (space group Cmmm) and the lattice parameters are a = 4.1450(1) Å, b = 12.3004(4) Å and c = 8.720(1) Å; atomic coordinates are given.

Introduction and literature review

The system Al–Cu has been investigated intensively during the last decades, mainly due to the importance of Al-based alloys, for example in aviation and transport industry. In view of this, the main focus of most studies in the system is the very Al-rich part. Although a lot of work was done in the Cu-rich part as well, there are still some uncertainties and inconsistencies in the phase diagram present.

The major assessment of the system was done in 1985 by Murray [1]. His extensive paper gives an equilibrium phase diagram as well as manifold information on metastable phase equilibria which are not part of the current investigation. According to [1], the equilibrium phase diagram contains 5 intermetallic compounds stable at ambient temperature and 7 additional compounds stable at elevated temperature (see Fig. 1). The phase diagram given by Murray does not represent the current level of knowledge about the system. A more recent phase diagram combining the assessment of Murray [1] with new data from Liu et al. [2] is given by Riani et al. [3]. Thermodynamic calculations in the system were performed by several authors concentrating on transition- and ordering phenomena [4], [5], [6], [7], as well as on atomic mobility [8]. A thermodynamic assessment is given by Saunders [9]. An overview on the Al–Cu phases described by different authors [1], [10], [11], [12], [13], [14], [15], [16], [17], [18], [19], [20], [21] is given in Table 1.

Fig. 1

The Al–Cu phase diagram according to Murray [1].

Table 1

Structural information on the compounds in the system Al–Cu.

Phase	Composition range [1]	Peason symbol	Space group	Structure type	Lattice parameters {Å}	Reference
(Al)	0–2.48	cF4	Fm-3m	Cu	a = 4.049750(15)	[17]
θ	31.9–33.0	tI12	I4/mcm	Al2Cu	a = 6.063(3)	[18]
					c = 4.872(3)
η1	49.8–52.4	oP16 or oC16	Pban or Cmmm	unknown	a = 4.087	[10][12]
					b = 12.00
		o∗32			c = 8.635
η2	49.8–52.3	mC20	C2/m	AlCu	a = 12.066	[11]
					b = 4.105
					c = 6.913
					β = 55.04°
ζ1	55.2–59.8	hP42	P6/mmm	Al3Cu4	–	[10]
					a = 8.1267(3)
		oF88	Fmm2		b = 14.4985(5)	[14]
					c = 9.9928(3)
ζ2	55.2–56.3	oI24–3.5	Imm2	Al3Cu4-δ	a = 4.0972(1)	[13]
					b = 7.0313(2)
					c = 9.9793(3)
ɛ1	59.4–62.1	Cubic?	unknown			[20]
ɛ2	55.0–61.1	hP4	P63/mmc	NiAs	a = 4.146(1)	[11]
					c = 5.063(3)
δ	59.3–61.9	hR52	R3m	Al4Cu9 (r)	a = 8.7066(1)	[15]
					α = 89.74(1)°a
γ0	59.8–69		I−43m	Cu5Zn8	–b	[2]
γ1	52.5–59	cP52	P−43m	Al4Cu9	a = 8.7068(3)	[16]
β0	67.6–70.2	unknown	unknown
β	70.6–82.0	cI2	Im-3m	W	a = 2.9504(2)	[19]
α2	76.5–78			long-period super structure based on Al3Ti and Cu3Au		[1]
(Cu)	80.3100	cF4	Fm-3m	Cu	a = 3.61491	[21]

* Bravais lattice is not known; ? Cubic symmetry is questionable.

Rhombohedral lattice parameters are given in non-standard setting for better comparison with cubic γ0.

No lattice parameters given.

The structure of the θ-phase with the composition Al2Cu was originally revealed by Friauf [22] and found to be tetragonal. According to Murray [1] the phase is stable up to 591 °C. Additional investigations in the region between 31 and 37.5 at.% Cu by Goedecke and Sommer [23] indicate a composition of 32.4 at.% Cu for θ at its formation temperature of 592 °C. The eutectic line of the reaction L = (Al) + θ ends at 32.05 at.% Cu. The widest solubility range of the phase is 0.55 at.% at 549 °C [23].

Structural investigations of the compound η-AlCu have already been performed by Preston [10] who found an orthorhombic structure in a sample quenched from 602 °C. Bradley et al. [20] investigated slowly cooled samples of the same composition and proposed an allotropic transformation η1→ η2 on basis of structural differences compared to the work of Preston. The authors suggested orthorhombic or monoclinic symmetry for the low temperature phase. El-Boragy et al. [11] were able to solve the structure of the low temperature phase which was found to be monoclinic. The high temperature structure is still unknown. Preston suggested the structure to be orthorhombic (oP16 or oC16) [10], Lukas and Lebrun [12] mentioned in their assessment of the Al–Cu–Si system an orthorhombic cell with lattice parameters a = 4.087 Å, b = 12.00 Å, c = 8.635 Å and 32 atoms per unit cell. Although the supposed type of transition reaction was not mentioned explicitly, the assessed phase diagram by Murray [1] and Riani et al. [3] obviously suggest a transition of higher order between η2 and η1.

According to the assessment of Murray, the introduction of the high temperature phase ɛ1 and ɛ2 goes back to 1920. However, the structure of ɛ2 was solved for the first time in 1972 by El-Boragy et al. [11], applying high temperature XRD. According to the authors, the structure of ɛ2-Al2+xCu3 is of the NiAs-type with partial occupation of the additional interstitial position. The structure of the high temperature modification ɛ1 is still unknown.

The compound with the proximate composition Al3Cu4 (ζ1/ζ2 – region) was also described by Preston [10] and Bradley [20] and was found to show a high and a low temperature modification. The work of Murray suggests a transition temperature between 530 and 570 °C but mentions other reported thermal effects between 373 and 450 °C as well [1]. Dong et al. [24], [25] investigated as-cast and annealed samples with the composition Al3Cu4. In the as-cast samples the authors find a mixture of an orthorhombic face-centered and an orthorhombic body-centered structure as well as a small amount of γ-Al4Cu9. After annealing at 500 °C for 10 h the oF structure became the major phase thus the authors suggested a transition Al4Cu9 + “oI” = “oF”. Electron Probe Micro Analysis (EPMA) measurements indicated compositions of Al43.2Cu56.8, Al41.3Cu58.7 and Al39.6Cu60.4 for “oF”, “oI” and γ-Al4Cu9, respectively. The crystal structures of ζ1 (Fmm2, structure type Al3Cu4) and ζ2 (Imm2, structure type Al3Cu4-δ) were finally solved by Gulay and Harbrecht using powder XRD [13], [14]. The composition of the samples for structure analysis of ζ1 (Al42.5Cu57.5) and ζ2 (Al43.2Cu56.8) contradicts the findings of Dong et al. [24], [25] who allocated the face-centered symmetry to the phase with lower Cu-content. Thermal analysis of samples by Gulay and Harbrecht [13], [14] reveals another contradiction. The assessment of Murray shows a low temperature phase ζ2 and a high temperature phase ζ1 with a slightly higher Cu-content; the transition temperature is supposed to be between 530 and 570 °C. Gulay and Harbrecht, however, found the Cu-richer phase ζ1 (only phase in a sample with the composition Al42.5Cu57.5) to be stable at 400 °C [14]. The Cu-poorer phase ζ2 (only phase in a sample with the composition Al43.2Cu56.8) was found to be stable at elevated temperatures (530 °C) and did not resist thermal treatment at 400 °C [13]. The authors claim that entropy provides an essential contribution to the stabilization of the ζ2 phase.

The range from 60 to 70 at % Cu was investigated intensively for many decades. Bradley [26] claimed that three different phases are present in this region: a cubic (γ), a monoclinic and a rhombohedral compound. Westman [27] found the latter to crystallize in space group R3m and confirmed the existence of a third phase of unknown structure between the cubic and the rhombohedral compound. Seshadri and Downie [28] claimed that there are only five intermetallic phases stable in the temperature range between 25 and 500 °C, namely γ(1), δ (described by a cubic structure), ζ2, η2 and θ. The separation of γ1 and δ is supported by an abrupt change in expansion coefficient from γ1 to δ. In the assessed phase diagram the third phase of unknown structure mentioned above is not included since there is no consensus about its existence. Murray pleads diffusion couple experiments of Funamizu et al. [29] which do not show any other phase between γ1 and δ. More importantly, very slow cooling experiments performed by van Sande et al. [30] show γ1 and δ in equilibrium, too. These two experiments support the non-existence of a third equilibrium phase in the indicated region. The existence of the high temperature phase γ0 was demonstrated by thermal analysis but the transition γ0 to γ1 could not be confirmed metallographically [1]. Liu et al. [2]. examined the Cu-rich part of the phase diagram by diffusion coupling, differential scanning calorimetry and high temperature XRD. The authors state that γ0 crystallizes in the Cu5Zn8-type and they do not find a two phase field between γ0 and γ1, thus proposing a higher order transition between the two phases, in contradiction to Murray’s assessment [1].

According to Murray [1], quoting Dawson [31], the high temperature phase β0 is formed peritectically from β and liquid at 1037 °C. Dawson determined the composition and stability range of β0 by metallography and dilatometry but the findings have never been reconfirmed and the structure of β0 remains unknown [1]. Nevertheless, β0 was included in the equilibrium phase diagram. In 1998 diffusion couple experiments performed by Liu et al. [2] showed a two phase region between β and γ0 and no single phase β0 was found. Additionally, the authors found only one peak in DSC measurements at 1019 °C which they interpreted to be the solidus of the β phase rather than the reaction temperature of the eutectoid transformation β0 = β + γ0. Hurtado et al. [32] investigated the region between 85 and 89 at.% Cu at temperatures from 450 to 850 °C, finding a square-like shaped phase which has, however, not been confirmed by other authors.

The β-phase and the two-phase region between β and (Cu) was frequently investigated and the assessment of Murray [1] gives a broad overview about the results of this research. It shows that the eutectoid temperature was found between 560 and 575 °C which can be explained by the sluggishness of the reaction. Reaction temperatures between 515 and 540 °C can be considered due to metastable eutectoid and peritectoid reactions.

The α2-phase was first described by West et al. [33] during long term annealing experiments. According to Murray, later studies confirmed the peritectoid decomposition temperature to be 363 °C at 77.25 at.% Cu. According to Murray’s assessment, α2 has an ordered fcc structure with a long-period superlattice based on Cu3Au and Al3Ti (Strukturbericht designations: L12 and D022, respectively) [1]. A more detailed description about investigations of the low temperature phase α2, including thermal analysis experiments in this region is given by Adorno et al. [34].

Experimental

The samples were prepared from Aluminum slug (99.999%), and Copper wire (99.95%), both supplied by Alfa Aesar, Karlsruhe, Germany. The Cu wire was reduced in a H2-flow at 300 °C for 3 h. The calculated amounts of Al and Cu were weighted to an accuracy of 0.05 mg; the sample weight usually was 1000 mg. Sample homogenization was done in an arc furnace MAM-1 by Edmund Buehler with a water-cooled copper plate and zirconium as the getter material. For homogenization of the sample, the resulting bead was turned and re-melted two times. The occurring mass loss during this procedure was found to be below 1% and therefore considered not to affect the sample composition significantly. The resulting bead was wrapped in Molybdenum foil (99.97%, Plansee SE, Reutte, Austria) and annealed at 500 °C under vacuum in a quartz glass tube for 24 days. Subsequently the samples were quenched in cold water and prepared for further investigation. Representative sections of all annealed samples were investigated by means of optical microscopy using a Zeiss Axiotech 100 microscope. Selected samples were analyzed by means of Scanning Electron Microscopy (SEM). The quantitative chemical analyses were performed on a Zeiss Supra 55 VP in combination with energy dispersive spectroscopy (EDS) using the pure elements for calibration. Measurements of the phase composition were performed with a minimum of three different spots and the results were averaged.

X-ray powder diffraction analyses were performed using a Bruker D8 ADVANCE diffractometer operating in reflection mode (Cu Kα1 radiation, LynxEye silicon strip detector). For selected samples high temperature X-ray powder diffraction analysis was applied using an Anton Paar XRK900 reactor chamber with an automated alignment stage. The temperature resolved measurements were performed under evacuated conditions. For evaluation of the resulting diffractograms, both at ambient as well as at elevated temperature, the software TOPAS [35] was used.

The DTA measurements were performed on a Setaram Setsys Evolution 2400 (Setaram Instrumentation, Caluire, France) and a Netzsch DTA 404 PC (Netzsch, Selb, Germany). The measurement devices used Pt/Pt-10%Rh thermocouples (Type S) which were calibrated using the melting points of pure Sn, Au and Ni. The samples with a weight of approximately 20 mg were placed in open alumina crucibles and measured under an argon flow of 50 ml min−1 for the Netzsch and 20 ml min−1 for the Setaram device. Applying a heating/cooling rate of 5 K·min−1, two consecutive curves were recorded for each sample. The possible mass loss during the DTA investigations was checked routinely and no relevant mass changes were observed.

Results and discussion

Phase equilibria in the system Al–Cu

Combining results from both DTA and SEM measurements, it was possible to obtain the complete description of the Al–Cu system which is plotted in Fig. 2. It is, for the most part, in good agreement to the phase diagram of Liu et al. [2] and Riani et al. [3] but it shows significant differences to the phase diagram proposed by Murray [1]. The occurring invariant reactions together with the composition of the reacting phases and the reaction temperature are given in Table 2, selected SEM images taken in back-scattered electron (BSE) mode are shown in Fig. 3.

Fig. 2

The Al–Cu phase diagram determined in the present work with experimental data points.

Table 2

Invariant reactions in the system Al–Cu according to the present work (bold) compared to Murray [1].

Reaction	Composition			Temperature (°C)	Ref.
L = (Cu)	–	100	–	1084.87	[1]
L = (Cu) + β	83.0	84.4	82.0	1032	[1]
	83.0(5)	84.5(5)	82.0(5)	1035(5)	this work
β = (Cu) + γ1	76.1	80.3	69	567	[1]
	76.0(5)	81.5(5)	70.0(5)	567(2)	this work
L + β = β0	69.2	70.9	70.2	1037	[1]
	reaction not confirmed				this work
β0 = β + γ0	70.0	70.6	68.5	964	[1]
	reaction not confirmed				this work
γ1 + (Cu) = α2	69	80.3	77.25	363	[1]
L + β0 = γ0	66.1	67.6	67.4	1022	[1]
	reaction not confirmed				this work
L = β	–	75	–	1049	[1]
				1052(5)	this work
γ0 = β + γ1	∼69	72.8	∼69	780	[1]
	reaction not confirmed				this work
γ0 = γ1		69.0		∼800	this work
		65.0		874(2)	this work
β + L = γ0	69.0(5)	63.0(5)	65.0(1)	993(2)	this work
γ0 + L = ɛ1	62.9	59.8	62.1	958	[1]
	65.5(5)	60.0(5)	64.5(5)	960(2)	this work
γ0 + ɛ1 = γ1	66.0	61.4	63.9	873	[1]
	reaction not confirmed				this work
γ1 + ɛ2 = δ	62.8	59.2	61.9	686	[1]
	63.0(5)	58.5(5)	61.5(5)	684(1)	this work
γ1 + ɛ1 = ɛ2	62.5	∼61.1	∼61.1	850	[1]
	64.0(5)	62.5(5)	62.5(5)	847(1)	this work
ɛ1 = ɛ2 + L	∼59.4	∼59.4	52.2	848	[1]
	59.5(5)	59.5(5)	52.5(5)	847(1)	this work
ɛ2 + L = η1	55.0	36.3	51.8	624	[1]
	54.5(5)	38.5(5)	52.0(5)	625(2)	this work
η1 + L = θ	59.8	32.2	32.8	591	[1]
	51.5(5)	32.5(5)	33.5(5)	591(2)	this work
ɛ2 = δ + ζ1	57.9	59.3	56.9	560	[1]
	reaction not confirmed				this work
ɛ2 = δ + ζ2	57.5(5)	60.0(5)	56.0(5)	578(2)	this work
ζ1 = ζ2 + δ	∼59.8	56.3	∼59.8	530	[1]
	reaction not confirmed				this work
δ + ζ2 = ζ1	60.0(5)	56.5(5)	57.0(5)	561(2)	this work
ζ1 + η1 = ζ2	55.2	52.3	55.2	570	[1]
	reaction not confirmed				this work
L = θ + (Al)	17.1	31.9	2.48	548.2	[1]
	17(1)	32.0(5)	2.5(5)	550(2)	this work
η1 = η2 + θ	49.8	49.8	33.0	563	[1]
	52.0(5)	52.5(5)	33.5(5)	574(3)	this work
ɛ2 + η1 = ζ1	56.5	52.4	56.2	590	[1]
	reaction not confirmed				this work
ɛ2 + η1 = ζ2	56.5(5)	53.0(4)	55.5(5)	597(1)	this work
ζ2 + η1 = η2	54.5(5)	52.5(5)	53.5(5)	580(1)	this work
η1 = η2 + ζ1	∼52.3	∼52.3	55.25	560	[1]
	reaction not confirmed				this work
L = (Al)	–	0	–	660.452	[1]

Fig. 3

BSE pictures of samples with the nominal composition A: Al90Cu10 [(Al) + θ], B: Al52.5Cu47.5 [θ + η2], C: Al42.5Cu57.5 [ζ1 + δ (traces)] and D: Al25Cu75 [γ1 + (Cu)].

The Al-rich part of the phase diagram has not been investigated extensively in the current study. Concerning the reaction temperatures and the phase composition, SEM and DTA measurements confirm the findings of previous authors. The solubility of Cu in Al was found to be 2.2(1) at.% at 500 °C. The lattice parameters of the phase θ (tI12, Al2Cu-type) vary from a = 6.0718(1) Å and c = 4.8802(1) Å at 32(1) at.% Cu to a = 6.0613(1) Å and c = 4.8724(1) Å at 33.6(2) at.% Cu. The phase boundaries of the binary phases in the Al-rich part of the phase diagram have been determined by means of SEM measurements and are indicated as black dots in Fig. 2.

In the region of a Cu-content higher than 50 at.%, the evaluation of phase equilibria was more challenging. The phases η1-AlCu and η2-AlCu are supposed to be stable around 50 at% Cu with a solubility range of 1–2 at.% Cu [1]. The results in the current investigations, however, reveal a shift of the composition toward the Cu-rich side. The solubility limits of the phase η2-AlCu at 500 °C were confirmed by SEM measurements. The Cu-poor and the Cu-rich composition limit were found at 51.9(5) at.% Cu and at 54.8(5) at.%, respectively. XRD analysis of the samples with the nominal composition Al49Cu51 (showing θ and η2 in equilibrium), Al47.5Cu52.5 and Al46.5Cu53.5 (showing single phase η2) and Al45Cu55 (showing η2 plus traces of ζ1) indicate a Cu-rich solubility limit between 54 and 55 at.% Cu which supports the SEM measurements.

The lattice parameters for η2 (mC20, AlCu-type) range from a = 12.0925(1) Å, b = 4.1001(1), c = 6.9085(1) Å, β = 55.03(1)° on the Cu-poor side to a = 12.2012(1) Å, b = 4.0997(2) Å, c = 7.0047(3) Å and β = 54.787(1)° for the Cu-rich side of this compound.

The phase diagrams of Riani and Murry indicate a transition temperature from the high temperature η1-phase to the low temperature η2-phase at 563 °C at the Cu-poor side and 560 °C at the Cu-rich side [1], [3]. According to our measurements the transition temperatures are 574(3) °C for the Cu-poor side and 580(1) °C for the Cu-rich side. Consequently, we propose also different solid state reactions in this area: the eutectoid decomposition η1 = η2 + θ at 574(3) °C and the peritectoid reaction η1 + ζ2 = η2 at 581(1) °C. The peritectic decomposition temperature of the phase η1 was found to be 625(2) °C which is in agreement with the phase diagrams mentioned above.

The region of Al4Cu9 with the supposed high temperature modification ζ1 and the low temperature phase ζ2 is also complex. Gulay and Harbrecht find the Cu-rich phase ζ1 stable at 530 and 400 °C, but the Cu-poor phase ζ2 stable at 530 °C and not stable at 400 °C [13], [14]. They discovered that after a heat treatment at 400 °C, ζ2 is decomposed into ζ1 and small amounts of η2. Additionally, Gulay and Harbrecht state that at 450 °C the phase ζ2 segregates in a mixture of ζ1 and ζ2. Therefore, the authors conclude that the temperature of the eutectoid decomposition of ζ2 is between 400 and 450 °C. The present work shows a sample with the nominal composition of Al45Cu55 exhibiting η2 as major phase with very small traces of a second phase. Comparison of systematic extinctions of ζ1 and ζ2 suggests that the second phase is ζ1. Due to the very low amount of ζ1 in the respective sample, SEM analysis of this sample shows only one suitable measurement point at 56.3 at.% Cu which can be assigned to the Cu-poor solubility limit of ζ1. XRD analysis of a sample with the nominal composition Al42.5Cu57.5 shows ζ1 [14] and some unidentified peaks (see Table 3). These XRD results narrow the solubility limits of ζ1 (oF88, Al3Cu4-type) between approx. 56 and 57.5 at % Cu. In the present investigation, DTA analysis of the sample with the nominal composition of Al45Cu55 does not show any effects related to the transition of ζ1 to ζ2, caused by the fact that the amount of ζ1 is very small in the sample. DTA analysis of the sample with the nominal composition Al42.5Cu57.5 shows an invariant effect at 561(2) °C which is considered to be related to the reaction ζ2 + δ = ζ1. These results do confirm the previous authors [13], [14] concerning the stability of the low temperature phase ζ1 which is considered to be stable between ambient temperature and 561(2) °C. The phase ζ2 is stable up to 596(1) °C where it decomposes peritectically. Since the XRD results of the sample Al45Cu55 show small traces of ζ1 we suggest a transition temperature ζ2 = ζ1 + η2 above 500 °C.

Table 3

Structure refinement of η1-Al1−δCu.

Compound	Al1−δCu
Number of formula units per unit cell	15
Space group	Cmmm
a {Å}	4.1450(1)
b {Å}	12.3004(4)
c {Å}	8.720(1)
Cell Volume {Å3}	444.53(3)
Number of atoms in the cell	30
Calculated density (g/cm3)	4.93(1)
Diffractometer	Bruker AXS D8-Advance
Radiation, wavelength {Å}	Cu Kα, 1.5406
Peak shape function	Fundamental parameter approach
Number of refined parameters	32
Rwp/GOF	3.88/1.34
Texture	Spherical harmonics, 4th order

The solubility limit at the Cu-rich side of ζ1/ζ2 was not accessible due to low contrast and fine microstructure. However, δ was found as the only phase present in Al40Cu60, while Al42.5Cu57.5 showed ζ1 with traces of δ. Therefore, the situation of the two phase field can be specified quite accurately.

It was not possible to determine the phase boundaries between the two phases δ and γ1 by SEM measurements due to the lack of contrast and possibly very fine microstructure. In XRD it was possible to distinguish the single phase region δ (R3m, a = 12.285(1) Å, c = 15.1486(1) Å [15]) from the single phase region γ1 (P-43m, a = 8.7068(3) Å [16]) by peak splitting and selective peak broadening even though the patterns look very similar. Samples with the nominal composition Al40Cu60 to Al37Cu63 show δ as the only present phase. The sample Al36Cu64 shows a main pattern corresponding to δ plus some small extra peaks that could not be explained by δ or γ1. They may be caused by super structure reflections, corresponding to the monoclinic structure proposed by Bradley [26] and Westman [27], which was omitted in the assessment of Murray [1]. We marked the respective area in Fig. 2 with a question mark. More detailed studies would be required to confirm the existence of an additional phase. The absence of invariant effects at 684(1) °C dedicated to the reaction ɛ2 + γ1 = δ in the sample Al38Cu62 and Al37Cu63 can be explained by a shift of the γ1 phase field toward the Cu-poorer region at elevated temperatures and, therefore, a smaller amount of δ taking part in the reaction. This leads to a smaller endothermic effect and since the respective endothermic peaks in the samples Al40Cu60 and Al39Cu61 are already small the resulting effect in the samples Al38Cu62 and Al37Cu63 might be insufficient to observe.

The region between γ1 and (Cu) has been the subject of an intensive research in the past and the present work does not provide any contradictory information. The solubility of Al in (Cu) as well as the upper solubility limit of γ1 was confirmed by SEM measurements. DTA analysis of samples in the respective area show very small thermal effects close to the solubility limit of (Cu) at 567 °C and thermal effects related to the formation of β at higher temperature, which are indicated as diamond shaped points in Fig. 2.

In general, solubility ranges and thermal stability of the high temperature compounds ɛ1 and ɛ2 could be confirmed in the present work by DTA observations. Slight changes concerning reaction temperatures and solubility ranges are shown in Table 2.

Since there is no consensus in literature concerning the transition of the high temperature phase γ0 to the low temperature form γ1 this area is of special interest. Analysis of thermal effects of samples in the respective field show very weak effects varying continuously with the composition. We did not observe any pointer for an invariant decomposition of γ1 in any of the investigated samples. Therefore we conclude that the transition γ1 = γ0 is of higher order, in agreement with the previous obtained results by Liu et al. [2].

Structural analysis of a sample with the nominal composition Al32Cu68 confirm the structure given for γ0 by Liu (I-43m, Cu5Zn8-type [2]), and reveals a lattice parameter of a = 8.8692(1) Å at 900 °C. XRD data of the sample at selected temperatures are shown in Fig. 4.

Fig. 4

High-temperature X-Ray powder diffraction of a sample with the nominal composition Al32Cu68. 25 °C: γ1, 750 °C: γ1, 900 °C: γ0 and traces of β.

According to Murray [1], the high temperature phase β0 was included in the equilibrium phase diagram although its existence could not be confirmed. Liu et al. [2] did not find any evidence of its existence and consequently was not incorporated in the assessed phase diagram of Riani et al. [3]. Results of the present work confirm that there is no evidence of the existence of β0 and all observed DTA effects in this composition area can be explained by the formation of the phases β and γ0.

Structural analysis of η1

Up to now, the crystal structure of η1 was not known. Preston [10] suggested the space groups Cmmm or Pban and orthorhombic lattice parameters were given by Lukas and Lebrun [12]. In the current study, we used high temperature powder XRD data to establish a structural model for η1. Measurements were carried out in a temperature range from 500 to 750 °C at intervals of 25 °C using a sample with the nominal composition Al50Cu50. A selection of these diffractograms is shown in Fig. 5. The measurements up to 550 °C show the low-temperature η2-phase in equilibrium with traces of the θ-phase. At 575 °C a third pattern, η1, appears which is the only phase present at 600 and 625 °C. Above 650 °C only ɛ2 is present and significant peak broadening can be observed, indicating the partial melting of the powder. After cooling back the sample to 500 °C one again observes η2 and θ, the diffraction lines, however, are significantly broadened. Some unidentified peaks of very low intensity may be attributed to oxide formation at the sample surface during the long stay in the non ambient device. Since the amount of possible oxide is very small, further investigation was not performed.

Fig. 5

High-temperature X-Ray powder diffraction of a sample with the nominal composition Al50Cu50. 500 °C: η1 and traces of θ, 600 °C: η1, 750 °C: ɛ2, 500 °C: η2, traces of θ and an unidentified peak (x) may be due to oxidation. Note that the diffractograms were recorded in the 2θ-range between 10 and 120°, but are only shown up to 44 °C for the sake of clarity of representation.

The pattern measured at 600 °C could be successfully indexed using the orthorhombic unit cell suggested by Lukas and Lebrun [12]. Cell refinement in space group Cmmm yielded the lattice parameters a = 4.1450(1), b = 12.3004(4) and c = 8.720(1) Å. According to the phase diagram discussion at 600 °C the measured sample with the nominal composition Al50Cu50 is in equilibrium with a small amount of liquid phase. This affects the background of the XRD measurement, which was compensated by modeling an additional broad peak at 43.9(1) °2θ. Further details of measurement and structure refinement of the η1 phase including the calculated errors of the parameters are listed in Table 3. The refined pattern of η1 is shown in Fig. 6.

Fig. 6

Refined powder XRD pattern of a sample with the nominal composition Al50Cu50 at 600 °C showing single phase η1.

The structural model for η1 was established by a twofold approach. Given the similarities of the lattice parameters between the monoclinic phase η2 and the orthorhombic η1, a ≈ b, b ≈ a and V ≈ 1.5∙V, we tried to develop the structural model by rearranging the atomic positions of the low temperature phase in the orthorhombic high temperature cell using the space group C222. This approach was supported by simulated annealing calculations [36] using the TOPAS software [35], [36]. The atomic coordinates of this structural model were finally transformed to Cmmm and standardized by applying the program Structure Tidy [37], [38]. During the consecutive Rietveld refinement, unusually large differences at the individual isotropic displacement factors were observed, indicating a decrease of electron density at some atomic positions. Therefore, all occupation factors were refined independently. The occupations of Al2, Al3 and Cu2 were found to be significantly reduced while all other sites were found to be fully occupied within 3 esd’s and were therefore fixed during the final refinements. The final structural model shows reasonable displacement factors and the refined overall composition Al14.1Cu14.8 (equivalent to 51.2 at.% Cu) is in excellent agreement with the Al-rich phase boundary of η1 (51.5(5) at.% Cu). The structural parameters of η1 are listed in Table 4. More details on the crystal structure investigation can be obtained from the Fachinformationszentrum Karlsruhe, 76344 Eggenstein-Leopoldshafen, Germany, (Fax: +497247 808666; e-mail: crysdata@fiz.karlsruhe.de) on quoting the depository number CSD 423053.

Table 4

Atomic coordinates, site occupancies and displacement factors for η1-Al1−δCu.

Atom	Site multiplicity Wyckoff letter	x/a	y/b	z/c	Occ	Beqa
Cu1	8n	0	0.1828(3)	0.1514(4)	1	1.0(1)
Cu2	4j	0	0.152(1)	1/2	0.70(1)	2.1(3)
Cu3	2c	1/2	0	1/2	1	1.1(2)
Cu4	2a	0	0	0	1	1.4(2)
Al1	8n	0	0.336(1)	0.338(1)	1	0.9(2)
Al2	4k	0	0	0.304(2)	0.68(2)	1.0(3)
Al3	4i	0	0.375(1)	0	0.84(2)	1.0(3)

Isotropic displacement factor as defined by Fischer and Tillmanns [39].

The coordination polyhedra for the 7 independent positions of η1 are shown in Fig. 7. Apart from the high-symmetry positions Cu3 and Cu4, the coordination figures are quite irregular with coordination numbers between CN = 10 and CN = 14. Interatomic distances in the first coordination sphere are given in Table 5.

Fig. 7

The first coordination sphere of the different atom positions in η1. Black: Cu, White: Al.

Table 5

Relevant interatomic distances (Å) for η1-AlCu (3.6 Å coordination sphere).

Atoms		Distance	Coordination number
Cu1	1 Al1	2.4824	12
	2 Al3	2.5597
	1 Al2	2.6134
	2 Al1	2.6396
	1 Al3	2.7132
	1 Cu4	2.6073
	1 Cu1	2.6404
	2 Cu1	2.6512
	1 Cu2	3.0626
Cu2	4 Al1	2.5171	14
	2 Al2	2.5322
	2 Al1	2.6698
	2 Cu3	2.7916
	2 Cu1	3.0626
	2 Cu2	3.1785
Cu3	4 Al1	2.4692	12
	4 Al2	2.6851
	4 Cu2	2.7916
Cu4	4 Al3	2.5770	10
	2 Al2	2.6523
	4 Cu1	2.6073
Al1	1 Cu3	2.4692	13
	1 Cu1	2.4824
	2 Cu2	2.5171
	2 Cu1	2.6396
	1 Cu2	2.6698
	1 Al1	2.8413
	2 Al2	2.9080
	2 Al1	2.9583
	1 Al3	2.9790
Al2	2 Cu2	2.5322	12
	2 Cu1	2.6134
	1 Cu4	2.6523
	2 Cu3	2.6851
	4 Al1	2.9080
	1 Al2	3.4143
Al3	4 Cu1	2.5597	11
	2 Cu4	2.5770
	2 Cu1	2.7132
	2 Al1	2.9790
	1 Al3	3.0631

A comparison of the atomic arrangements in the low temperature phase η2 and the high temperature phase η1 is shown in Fig. 8. The figure shows the layer in (001) of orthorhombic η1 in comparison to the layer in (010) of the monoclinic η2, i.e. both structures are projected along their short axis. All atoms shown are situated within the mirror plane at z = 0 and y = 0, respectively. The corresponding second layer of each structure (situated at z = 1/2 and y = 1/2, respectively) shows the same atomic arrangement shifted by ½ in [010] for η1 and in [100] for η2 according to space group symmetry.

Fig. 8

Comparison of the (100)-plane of the high temperature phase η1 and the (010)-plane of the low temperature phase η2. Black: Cu, White: Al.

Fig. 8 shows that both structures have a common structural motif; i.e. a diamond shaped unit consisting of 5 Cu- and 4 Al-atoms. These motifs are arranged in a rectangular pattern and interconnected along their corners in case of the high temperature structure. In the monoclinic structure the motifs are re-arranged and interconnected diagonally along their edges. While only one atom, Al3, is not part of the diamond-shaped motif of η2, three of the seven sites in η1 (Al1, Cu2 and Cu3) are not part of this motif.

Although the two structures are obviously related it should be pointed out that it is not possible to transform one structure into the other in a simple way and a second order transition between η1 and η2 can be definitely ruled out. This is consistent with our phase diagram investigation which clearly indicates an invariant reaction related to the transformation from η2 to η1. As an example, the DTA curve for the sample with the nominal composition Al50Cu50 (used also for the high temperature XRD) is shown in Fig. 9. At onset 571 °C a sharp reaction peak corresponding to the eutectoid formation of η1 occurs. This effect is followed by the peritectic decomposition of θ (590 °C) and η1 (625 °C) and finally the liquidus effect at 804 °C.

Fig. 9

DTA curve for the sample with the nominal composition Al50Cu50, showing thermal effects related to the η1/η2-transition.

Summary

The current work revealed significant improvements on the established phase diagram in the system Al–Cu by solving several inconsistencies in literature. The Al-rich part of the phase diagram could be confirmed. The phases η1 and η2 show a significant shift to the Cu-rich side of the phase diagram and exhibit, contrary to previously published phase diagrams [1], [2], [3], a first order transition reaction. The ζ1/ζ2 region was re-investigated and the recent findings of Gulay and Harbrecht [13], [14] were widely confirmed. The transition between γ0 and γ1 does not show a two phase field thus indicating a higher order transition and confirming the results of Liu et al. [2]. The absence of the high temperature phase β0 was confirmed.

The structure of the high temperature phase η1 was determined from powder diffraction data. The phase is orthorhombic (space group Cmmm) and the lattice parameters are a = 4.1450(1) Å, b = 12.3004(4) Å and c = 8.720(1). The structural relations to the low-temperature compound η2 are discussed.