Phase equilibria in the neodymium-cadmium binary system.

The equilibrium phase diagram of the neodymium-cadmium system has been established by thermal, metallographic and X-ray analysis based on a study of 70 alloys. The system contains three congruently melting intermetallic compounds, i.e. NdCd (1040 °C), NdCd2 (995 °C), Nd11Cd45 (855 °C), and four incongruently melting compounds NdCd3 (860 °C), Nd13Cd58 (740 °C), NdCd6 (655 °C) and NdCd11 (520 °C). Four eutectic reactions are found in this binary system, i.e. at ∼25 at.% Cd and 770 °C, at 58 at.% Cd and 955 °C, at 79 at.% Cd and 850 °C, and very close to pure Cd at 318 °C, as well as one eutectoid reaction at ∼15 at.% Cd and 500 °C. The solid solubility of Nd in Cd is negligible. Dilatometric curves were recorded for three Nd-Cd compositions up to 4 at.% Cd, to accurately determine phase transitions between the solid solutions of Cd in the low- and high-temperature modification of Nd.

Introduction

Pyrochemical reprocessing techniques for high burn-up nuclear fuels promise more efficiency than traditional aqueous methods [1,2]. As described by Olander [2], pyrochemical reprocessing is based on an “electro-refining” process where a liquid metal cathode is applied for the selective recovery of Pu and minor actinides (MA). At least three different low-melting metals were investigated for this purpose: Kurata et al. [3] proposed Bi and Cd, and later on Conocar et al. obtained promising results using liquid Al [1]. As rare earth elements (RE) behave chemically in a similar way compared to minor actinides, optimal reprocessing conditions require also a detailed knowledge on the extractability of these elements, as described by Kurata et al. [3]. Accordingly, the formation of RE-Me intermetallic compounds affects the performance of the extraction process. For its optimization and better efficiency the knowledge on the existence and stability of the occurring compounds in the various binary phase diagrams RE-Me is important.

The present study has been carried out as part of a research project on the interaction of different RE elements (Ce, Pr, Nd and Gd) with Cd [4-7]. It presents the phase equilibria in the Nd–Cd system as a function of temperature and composition and specifies the existing Nd–Cd intermetallic compounds as well as all phase reactions among them.

A first partial Nd–Cd phase diagram was reported by Elliot [8] who presented the Cd-rich part for temperatures up to 600 °C. The first more detailed study of the Nd–Cd phase diagram was provided by Iandelli [9,10] who described the crystal structure of four intermetallic compounds, i.e. NdCd, NdCd2, NdCd3 and NdCd11, applying X-ray diffraction (XRD). Nevertheless, no information concerning the melting temperatures of these compounds was given. It was later summarized by Gschneidner [11] with additional information about two allotropic modifications of Nd: α-Nd [12] and the high temperature modification β-Nd [13]. Further work in this system was done by Johnson et al. [14,15] who determined the solubility of Nd in liquid Cd by chemical analysis of filtered samples of the equilibrium liquid phase in the temperature range 347–498 °C. The most Cd-rich compound in equilibrium with liquid Cd is NdCd11. The peritectic decomposition of this phase at 530 °C was examined for a sample containing about 5 at.% Nd using differential thermal analysis (DTA). A maximum solubility of Nd at this peritectic was calculated as 2.6 at.% [15,16]. The existence of cubic RECd6 was discussed by Johnson et al. [16], and later on the crystal structure of YCd6, that shows close relationship with NdCd6, was determined [17]. Bruzzone et al. [18] determined the existence and crystal structure of the following Nd–Cd compounds by use of metallographic and X-ray methods: NdCd4 (γ-brass type) and Nd2Cd9∼ (Pu13Zn58 type). The existence of a phase with the Pu13Zn58 type structure had been reported earlier by Iandelli and Ferro [19] but only for the La–Cd and Ce–Cd system. A rudimentary phase diagram of the Nd–Cd system can be found in Massalski’s handbook [20], based mainly on the earlier work by Johnson et al. [15,16] and on the assessment of Gschneidner and Calderwood [21]. Seven intermetallic compounds are known which were also summarized by Colinet and Pasturel [22]. Table 1 compiles crystallographic data for all binary compounds and solid solutions that were determined in the Nd–Cd phase diagram.

Table 1

Crystal structure data (from literature) of phases in the Nd–Cd system.

Phase	Composition (at.% Cd)	Structure type	Pearson symbol	Space group	Refs.
α-Nd	0–3.0a	α-La	hP4	P63/mmc	[12]
β-Nd	0–19.0a	W	cI2	Im3m	[13]
NdCd	50.0b	CsCl	cP2	Pm3¯m	[9,10]
NdCd2	66.7b	CdI2	hP3	P3¯m1	[9,10]
NdCd3	75.0b	BiF3	cF16	Fm3¯m	[9,10]
Nd11Cd45	80.4b	γ-brass	cF448	F4¯3m	[18]
Nd13Cd58	81.7b	Pu13Zn58	hP142	P63/mmc	[18]
NdCd6	85.7b	YCd6	cI168	Im3¯	[17]
NdCd11	91.7b	BaHg11	cP36	Pm3¯m	[9,10]

Derived from the present results.

Given by Massalski [20].

Veleckis and van Deventer [23] listed temperatures for various “RE-RECd eutectics” obtained experimentally, among them a value of 486 °C for the Nd–Cd system. However, this value is obviously flawed due to the approximation of results derived from the Denbigh equation [21] and is far off the value proposed later in Ref. [20] (about 650 °C).

Experimental

The starting materials used to prepare the Nd–Cd alloys were Cd shot (United Mineral & Chemical Corporation, U.S.A., and Alfa Aesar, Karlsruhe, Germany; both 99.9999%) and Nd in form of foil/rod (Alfa Aesar, Karlsruhe, Germany; 99.9%) and pieces (smart-elements, Vienna, Austria, 99.9%). Due to the high reactivity of Nd the preparation procedure was done in a glove box (MBraun Labmaster 130) under an Ar atmosphere (<1 ppm O2, <1 ppm H2O). Before use, the Nd pieces were cleaned with a file until a shiny surface was obtained. Calculated amounts of Nd and Cd were weighed on a semi-micro balance with an accuracy of at least ±0.5 mg.

The high reactivity of Nd with water and oxygen [24] as well as the high volatility of Cd restricts the number of materials that can be used for sample preparation. RE metals attack alumina and quartz glass largely at high temperature preventing their application. Tantalum had been found to be an ideal material. No intermetallic compounds are formed in the Nd–Ta system [20], and only a very slight pick-up of Ta by RE metals was reported by Spedding and Daane [25]. Dennison et al. [26,27] determined the very low solubility of Ta in liquid Nd only for high temperatures between 1547 and 2070 °C. The result of the assessment by Garg et al. [28] indicated a negligible (0.006 at.%) solubility of Nd in Ta at 1021 °C. In addition, a Ta crucible can be easily closed with a corresponding Ta lid by welding in a standard arc furnace [29] providing sufficient protection of the contained sample.

Therefore, all alloys were prepared in Ta crucibles (10 mm O.D., 15 mm height) which were closed by welding with a corresponding lid in a water-cooled arc furnace under an Ar pressure of 0.5 bar. The mass loss during the whole sample preparation procedure was usually below ±0.2 mg and did not significantly affect the sample composition. The total sample weight was usually 1 g, for dilatometric studies 2 g samples were prepared. The closed Ta crucibles were sealed separately in quartz glass tubes under 10−3 mbar and placed into a muffle furnace for various durations ranging from 3 to 12 weeks. Initially, all samples were melted for proper homogenization for at least 3 days; for samples containing ⩽70 at.% Cd a temperature of 1050 °C was selected which is slightly above the melting point of pure Nd, all other samples were kept at 700 °C. Afterward, different heat treatment steps were applied in the temperature range between 200 and 850 °C (see Table 3). The annealed samples were quenched in cold water and characterized by X-ray diffraction (XRD), light optical microscopy (LOM), energy dispersive X-ray spectroscopy (EDS) and differential thermal analysis (DTA). Additional dilatometric measurements were carried out for three Nd-rich samples.

Table 3

Comparison of phase boundaries at different temperatures derived from SEM analyses and isopiestic vapor pressure measurements [30].

Compound	SEM		Isopiestic experiment
	Temperature (°C)	Phase boundaries (at.% Cd)	Temperature (°C)	Phase boundaries (at.% Cd)
α-Nd	500	0–3.0	–	–
β-Nd	770	0–19.0	–	–
NdCd	300	47.8–50.0	550	48.7–50.1
	600	47.4–50.2
	850	47.0–50.6
NdCd2	600	65.0–66.7	550	65.7–67.1
	850	65.5–66.7
NdCd3	600	74.4–75.0	–	–
	850	74.4–74.8
Nd11Cd45	600	79.2–80.5	550	79.7–80.3
Nd13Cd58	300	81.4–81.7	550	81.2–81.7
	600	81.5–81.8
NdCd6	300	85.0–85.7	550	84.6–85.7
	600	84.6–85.6
NdCd11	300	91.3–91.8	–	–

Phase identification was done by powder XRD using Cu Kα radiation on a Bruker D8 Advance Diffractometer with Bragg–Brentano focusing geometry. For this purpose, sample pieces were powdered inside the glove box and afterward glued to a carrier by use of Vaseline dissolved in n-hexane. Special sample holders with X-ray transparent lids were used to prevent sample powders from oxidation. The XRD patterns were analyzed and refined by means of the TOPAS 3 software (provided by Bruker), applying the fundamental parameter approach for peak profile modeling.

For metallographic studies, pieces of the quenched samples were first embedded in a resin/carbon mixture, ground under paraffin oil or kerosene, polished with an oil-based diamond suspension and finally cleaned in an ultrasonic bath with n-hexane. With increasing Nd content the surface oxidation rate increased, hence, the polishing of samples containing more than 50 at.% Nd was more laborious and required a rapid transfer into the glove box. The quality of the polished samples was checked in a LOM (Zeiss Axiotech 100). In order to investigate microstructures and chemical compositions of the samples, a scanning electron microscope (SEM: Zeiss Supra 55 VP) with directly coupled EDS was applied. The measurements were performed at an acceleration voltage in the range of 15–20 kV. Pure Co acted as a standard to control parameters of the measured beam.

Differential thermal analyses were carried out on a DTA 404 S and DSC 404 PC as well as on a SETARAM Setsys Evolution, using Pt/Pt10%Rh thermocouples. The instruments were calibrated at the melting points of high purity metals: Ag, Cu, Ni and Zn. A sample mass of 100 mg was used for the experiments and high-purity γ-Al2O3 was employed as reference material. All samples, except pure Cd, underwent two heating and cooling cycles using a heating/cooling rate of 2 K min−1 (as-cast samples) and 5 K min−1 (annealed samples). They were contained in hermetically closed Ta-crucibles with plane bottoms (for better contact) which were formed using a hydraulic press. A constant Ar flow of 50 ml min−1 was applied in the DTA furnace chamber. The accuracy of the temperatures determined from the DTA curves is estimated to be better than ±2 °C.

Samples containing 1, 2, and 4 at.% Cd were also studied on heating and cooling by dilatometry using a horizontal dilatometer (Netzsch 402E) with inner tube diameter of 35 mm. The samples were prepared in closed Ta tubes (12 mm O.D., 35 mm long) and annealed at 200 °C. After an equilibration time of 12 weeks, the alloys were cut and precisely machined into the form of cuboids. In order to compare the DTA results with those from dilatometry, similar experimental conditions like heating and cooling rate (5 K min−1) as well as an Ar flow of 50 ml min−1 were used.

Results and discussion

The equilibrium phase diagram

A total number of 70 samples were annealed and characterized by powder-XRD, SEM and DTA to obtain a complete description of the Nd–Cd phase diagram. Equilibrated samples and as-cast alloys were consulted to define homogeneity ranges, phase equilibria and crystallization behavior. Relevant samples, examined with isothermal methods, are listed in Table 2. Moreover, dilatometric measurements were performed to determine phase boundaries in the Nd-rich part (compare chapter 3.3.).

Table 2

Experimental phase compositions and lattice parameters of equilibrium phases in selected Nd–Cd samples.

Sample/nom. comp. (at.%)	Annealing temp. (°C)	Phase analysis		SEM
		Phases	Lattice parameters (Å)	Cd (at.%)	Nd (at.%)
1	300	Cd	a = 2.9794(5), c = 5.6194(1)	100	0.0
Nd3Cd97		NdCd11	a = 9.3011(5)	91.6	8.4
2	300	NdCd11	a = 9.2957(9)	91.3	8.7
Nd11Cd89		NdCd6	a = 15.6285(2)	85.7	14.3
3	300	NdCd6	a = 15.6318(1)	85.0	15.0
Nd16Cd84		Nd13Cd58	a = 15.6152(1), c = 15.4661(1)	81.6	18.4
4	600	Nd13Cd58	a = 15.6250(3), c = 15.4316(4)	–a	–a
Nd19Cd81		Nd11Cd45	Low intensity	80.5	19.5
5	600	Nd11Cd45	a = 21.7817(2)	79.2	20.8
Nd23Cd77		NdCd3	a = 7.1692(3)	75.0	25.0
6	600	NdCd3	a = 7.1792(4)	74.5	25.5
Nd30Cd70		NdCd2	a = 5.0232(2), c = 3.4489(2)	66.6	33.4
7	850	NdCd3	a = 7.1804(5)	75.3	24.7
Nd28Cd72		NdCd2	a = 5.0231(4), c = 3.4474(4)	–a	–a
8	600	NdCd2	a = 5.0255(4), c = 3.4440(3)	65.0	35.0
Nd46Cd54		NdCd	a = 3.8198(2)	50.6	49.4
9	850	NdCd2	a = 5.0228(8), c = 3.4518(8)	65.5	34.5
Nd37Cd63		NdCd	a = 3.8204(5)	50.6	49.4
10	600	NdCd	Too ductile	47.4	52.6
Nd70Cd30		β-Nd	Too ductile	16.5	83.5

The microstructure of the respective phase was too fine for an accurate measurement by EDX.

Based on the combined results from isothermal and dynamic methods as well as dilatometric measurements, a complete version of the Nd–Cd phase diagram was drawn, which is given in Fig. 1. All seven intermetallic compounds which were found in literature (Table 1) could be obtained in equilibrated samples. Lattice parameters and phase compositions of phases in relevant samples are listed along the complete composition range in Table 2. The corresponding phase boundaries of all intermetallic compounds as well as solid solutions are listed in Table 3 together with estimated homogeneity ranges from isopiestic vapor pressure measurements [30]. It is striking that the homogeneity ranges of all phases appear to be shifted toward the Nd-rich side of the nominal stoichiometry, something that had already been observed in other RE-Cd systems [4-7].

Fig. 1

The Nd–Cd phase diagram according to the present results.

All experiments to obtain single-phase samples of the solid solutions of Nd by quenching failed. Although the solid solution of Cd in α-Nd could be clearly identified from the present EDX results, minor amounts of unreacted Nd were always present. Similarly, the high-temperature modification β-Nd could be observed in quenched samples but most of the corresponding microstructures showed inhomogeneities.

Besides, XRD investigations showed the presence of unreacted Cd in some quenched samples with nominal compositions between 0 and 90 at.% Cd (cf. the pattern of sample Nd37Cd63 in Fig. 2). It has to be assumed that, due to rapid quenching of the sample, evaporated Cd condensed on the surface of the specimen. Therefore, the XRD pattern shows traces of Cd while the SEM micrograph does not. In fact, these relatively small amounts of Cd which had been evaporated and re-condensed on the sample surface on quenching did not significantly change the nominal compositions of the samples. Thus, the expected phase equilibria were reached in most of the samples which had been prepared with nominal compositions located in two-phase fields (see Table 2).

Fig. 2

The XRD pattern of the Nd37Cd63 sample annealed at 850 °C for 3 weeks.

While most of the samples which contained intermetallic compounds turned out to be rather brittle, powdering of alloys containing the phases NdCd and NdCd2 was very difficult due to their rather significant ductility. This is in agreement with observations in the related Gd–Cd system [31] where Vickers hardness measurements showed that the phases GdCd2 and GdCd exhibited the lowest hardness values. Since GdCd2 and GdCd are isotypic with the present compounds NdCd2 and NdCd a similar mechanical behavior was to be expected and, in fact, observed. Consequently, the XRD-powder patterns of NdCd2 and NdCd showed rather broad peak shapes. Additionally, absorption phenomena of these Nd-rich phases lowered also the intensities and made it altogether difficult to derive accurate lattice parameters.

Evaluation of isothermal and dynamic measurements

Samples studied by DTA are listed in Table 4, together with their thermal effects from two heating and cooling cycles, respectively. The temperatures for invariant effects in Fig. 1 are given as the average values from several independent measurements indicating error limits of ±5 °C or less. Generally, the results obtained for the as-cast samples are in good agreement with those for the annealed ones.

Table 4

Experimental results of DTA in the Nd–Cd system.

Sample comp. (at.%)	Annealing T (°C)/weeks	Heating (°C)						Cooling (°C)
		Invariant reactiona				Phase boundary	Liquidus	Liquidus
Nd99Cd1	250/12					325 705–777	n-o	n-o
Nd98Cd2	300/10					613–809	1016	1015
Nd95Cd5	as-cast	498				n-o	970	968
Nd95Cd5	600/4	495				n-o	967	961
Nd90Cd10	as-cast	495				n-o	923	922
Nd90Cd10	800/3	502				n-o	914	913
Nd88Cd12	400/6	504				n-o	893	892
Nd80Cd20	as-cast	495	775				813	813
Nd80Cd20	400/6	498	773				804	803
Nd78Cd22	400/6	503	770				n-o	787
Nd70Cd30	as-cast	493	771				941	n-o
Nd60Cd40	400/6	504	769				1010	1004
Nd55Cd45	as-cast	n-o	765				1039	1033
		500 ± 5	770 ± 5				1018	1012
Nd46Cd54	as-cast	954	959				1030	1021
Nd46Cd54	400/6	950	957				1030	1023
Nd37Cd63	as-cast	955	961				991	990
Nd37Cd63	600/4	955	963				993	991
		955 ± 5	960 ± 3
Nd35Cd65	600/4	n-o	866				992	965sc
Nd30Cd70	as-cast	864	n-o				964	950sc
Nd26Cd74	600/4	861	866				n-o	n-o
Nd23Cd77	as-cast	863	865	849			883	860sc
Nd23Cd77	600/4	860	865 ± 1	853			n-o	n-o
Nd22Cd78	600/4	859		852			n-o	n-o
		860 ± 4		850 ± 3
Nd19Cd81	as-cast			n-o			853	834sc
Nd16Cd84	as-cast			738			843	812sc
Nd16Cd84	600/4			740			834	834
Nd12Cd88	as-cast	515	653	744			775	750sc
				740 ± 4
Nd10Cd90	400/6	520	n-o				744	724sc
			655 ± 2
Nd7Cd93	as-cast	520			316		n-o	650sc
Nd5Cd95	300/10	523			320		n-o	620sc
Nd4Cd96	as-cast	521			316		622	577sc
		520 ± 3			318 ± 2

n-o not observed, sc supercooled effect.

Average values of all invariant reaction temperatures are given together with estimated error limits in bold numbers.

For each sample two heating and cooling cycles were performed at a heating rate of 2 K min−1 (as-cast samples) and 5 K min−1 (annealed samples) in order to check if equilibrium conditions can be restored after the first melting of the sample. Liquidus effects were evaluated both on heating and cooling if possible; all other effects were taken from the heating curves only. All invariant reactions, evaluated from the present results, are listed in Table 5 together with the respective reaction temperatures and types and the phase compositions. As can be seen, the intermetallic compounds NdCd11, NdCd6, Nd13Cd58 and NdCd3 are formed incongruently whereas Nd11Cd45, NdCd2 and NdCd show congruent melting behavior.

Table 5

Invariant reactions in the system Nd–Cd derived from a combination of all present results.

Reaction	T (°C)	Phase compositions (at.% Cd)			Reaction type
L ⇄ Cd + NdCd11	318 ± 2	∼100	∼100	91.7	Degenerate eutectic
L + NdCd6 ⇄ NdCd11	520 ± 3	97.4	85.9	91.7	Peritectic
L + Nd13Cd58 ⇄ NdCd6	655 ± 2	94.4	81.7	85.7	Peritectic
L + Nd11Cd45 ⇄ Nd13Cd58	740 ± 3	89.9	80.4	81.7	Peritectic
Nd11Cd45 ⇄ L	855		80.4		Congruent melting
L ⇄ Nd11Cd45 + NdCd3	850 ± 3	79.0	80.3	75.0	Eutectic
L + α-NdCd2 ⇄ NdCd3	860 ± 4	78.2	66.7	75.0	Peritectic
α-NdCd2 ⇄ β-NdCd2	865 ± 1		66.7		Polymorphic transformation
	955 ± 3		66.1
β-NdCd2 ⇄ L	995		66.7		Congruent melting
L ⇄ β-NdCd2 + NdCd	960 ± 3	58.1	66.1	50.5	Eutectic
NdCd ⇄ L	1040		50.0		Congruent melting
L ⇄ NdCd + β-Nd	770 ± 5	24.0	47.2	19.0	Eutectic
β-Nd ⇄ NdCd + α-Nd	500 ± 5	15.0	47.5	3.0	Eutectoid

According to literature [11], the transition between α-Nd and β-Nd occurs at 862 °C, and the melting point of Nd is at 1024 °C [11] or 1021 °C [21], respectively. Attempts to reproduce these values gave inconclusive results: the α to β transition appeared to be at lower temperature, and the melting point was indicated at higher temperature; however the corresponding melting effect in the DTA resembled more a typical non-invariant effect than a clear DTA signal which one might expect from an invariant melting point of a pure substance. From this it must be concluded that the impurity content (probably most of it oxygen) of the Nd was still high enough to prevent the determination of a reliable melting point. Thus we kept the literature values for the α-Nd to β-Nd transition (862 °C) and the melting point of β-Nd (1021 °C) for the present phase diagram.

Extended solid solutions of Cd in both α-Nd and β-Nd were reported earlier [12,13] and could be confirmed in the present study. The respective solubility limits of Cd within Nd were determined at about 3 (α-Nd) and 19 at.% (β-Nd) in the present study (cf. Table 1). It was observed that Cd stabilizes β-Nd down to 500 °C were it decomposes in a eutectoid reaction forming α-Nd and NdCd. At 770 °C β-Nd is formed according to the eutectic reaction L = β-Nd + NdCd. The recorded DTA curve for the sample Nd60Cd40, annealed at 400 °C for 6 weeks, is given in Fig. 3. Both, the eutectoid and the eutectic reaction as well as the melting effect can be clearly distinguished. Nevertheless, the eutectoid reaction could not be observed in the heating and cooling curves of the sample Nd55Cd45. Most probably, the effect is already too weak due to an insignificant amount of α-Nd in the sample.

Fig. 3

DTA curves of the sample Nd60Cd40, annealed at 400 °C for 6 weeks, measured upon heating and cooling with 5 K min−1.

The compound NdCd melts congruently at 1040 °C and forms a eutectic (58 at.% Cd) with NdCd2 at 960 °C: L = NdCd + NdCd2. The microstructure of the sample Nd46Cd54, representing the structure of the eutectic between NdCd and NdCd2, is shown in Fig. 4. NdCd shows a relatively broad homogeneity range. Its limiting compositions were determined as 47 and 50.6 at.% Cd at 850 °C (cf. Table 3).

Fig. 4

Microstructure of the sample Nd46Cd54 annealed at 600 °C for 4 weeks.

For compositions between 54 and 77 at.% Cd some additional thermal effects were noticed in the temperature range 860–960 °C, most probably corresponding to the transformation of NdCd2 into a high temperature modification (shown as β-NdCd2 in Fig. 1). Such a polymorphic modification of RECd2 intermetallic phases was already found in other RE-Cd systems, for example in the systems Gd–Cd [31] and Pr–Cd [4]. Unfortunately, all attempts to synthesize the high temperature modification β-NdCd2 and retain it by quenching were unsuccessful. The intermetallic compound NdCd2 has a congruent melting point at 995 °C and takes part in the peritectic formation reaction of NdCd3 at 860 °C which is clearly observed by DTA for samples between 70 and 78 at.% Cd. For example, for the sample Nd22Cd78 two sharp DTA peaks were registered: the peritectic decomposition of NdCd3 as well as the eutectic reaction between NdCd3 and Nd11Cd45 at 850 °C and 79 at.% Cd. No separate liquidus effect could be recognized in this sample, neither on heating nor on cooling, which indicates that it must be very close to the peritectic reaction temperature.

The intermetallic compound Nd11Cd45 with a nominal composition of 80.4 at.% Cd melts congruently at 855 °C which is an average value of several independent DTA measurements. Metallographic examination of the sample Nd22Cd78, annealed at 600 °C for 4 weeks, showed the presence of both phases, NdCd3 and Nd11Cd45 (Fig. 5). At this temperature Nd11Cd45 dissolves about 1 at.% Nd (cf. Table 3). This solubility was also indicated by isopiestic vapor pressure measurements [30].

Fig. 5

Microstructure of the sample Nd22Cd78 annealed at 600 °C for 4 weeks.

It can be recognized that the liquidus curve decreases rather steeply between 80.4 and 100 at.% Cd. In this composition range, three consecutive peritectic reactions occur: at 740 °C the phase Nd13Cd58 is formed, followed by the peritectic formation of NdCd6 and NdCd11 at 655 and 520 °C, respectively. NdCd6 showed a slight variation of the solubility of Nd with temperature and its homogeneity range is limited on the Nd-rich side at 84.6 at.% Cd at 600 °C. No significant solid solubility was determined for NdCd11 which seems to be a line compound.

Similarly as reported by Johnson et al. [15], the liquidus temperature of an annealed sample with 5 at.% Nd could not be determined from the heating curve, however, it was observed during cooling. Nevertheless, for the as-cast sample Nd4Cd96 it was clearly registered also on heating using a rate of 2 K min−1.

In other RE-Cd systems, Bruzzone and Merlo [32] and Canepa et al. [33] indicated the tendency to supercooling for compositions between 3 and 15 at.% La or Ce, respectively. A similar tendency was also observed in this work, especially for sample compositions up to approximately 35 at.% Nd (see Table 4).

The shape of the liquidus curve between 97.4 and 100 at.% Cd was drawn following the extrapolation of the data listed in Ref. [15]. A strong invariant effect was noticed at 318 °C, which corresponds to the eutectic L = NdCd11 + Cd. When extrapolating the equation, given by Johnson et al. [14], to the eutectic temperature, the eutectic point itself would be at 0.017 at.%. No significant solid solubility of Nd in Cd was detected by SEM in the present study.

For the Cd-rich part of the phase diagram, the detected reaction temperatures were roughly comparable with values given in Ref. [20]: the peritectic decomposition of NdCd11, reported at 530 °C in Ref. [20], was found here at 520 °C, and the eutectic between NdCd11 and Cd was found at 318 °C instead of 321 °C in Ref. [20]. The value of 318 °C seems to be reasonable if compared with the published experimental data for the analogous eutectics in the Ce–Cd [33] as well as the Gd–Cd [31] systems, i.e. 315 and 316 °C, respectively.

For the annealed DTA samples, oxidation of the Ta-crucibles was observed after DTA measurements in cases when samples where heated above 1000 °C. Apparently, Ta behaved as a getter material for oxygen traces in the DTA furnace chamber. Selected Ta crucibles were sectioned and both surface and cross-section were analyzed by SEM. Only minor amounts of Ta (up to 0.3 at.%) were observed in the corresponding samples, mostly in the part directly sticking to the crucible walls. It is assumed that this arises rather from a contamination during sample polishing than from a reaction of the sample with the crucible material itself.

Dilatometric study

Dilatometric curves display several effects observed upon heating of solid Nd–Cd samples. Although the signals of the recorded dimensional changes were weak, they nevertheless provided some additional information. Due to the high volatility of Cd only the following Nd-rich samples were investigated: Nd99Cd1, Nd98Cd2 and Nd96Cd4. The observed curve for the sample with 2 at.% Cd shows a similarity to the recorded DTA curve. It is presented in Fig. 6a, and the results can be summarized as follows:

Fig. 6

Dilatometric curve (thick line) together with temperature curve (thin line) for the samples Nd98Cd2 (a) and Nd96Cd4 (b), recorded with a heating rate of 5 K min−1.

a sharp initial contraction (curve A–B);

an expansion during heating with a slight maximum at about 440 °C (B′) corresponding to the phase boundary crossing α-Nd + NdCd → α-Nd (curve B–C);

a decrease of the expansion rate between about 750 and 815 °C representing probably the transition into the two-phase field α-Nd + β-Nd (curve C–D);

an increased expansion in the range between 815 °C and 950 °C (curve D–E);

a shrinkage during the isothermal holding at 950 °C (curve E–F);

a further contraction during cooling with a small effect (G) at about 810 °C which is probably connected to the phase boundary crossing β-Nd → α-Nd + β-Nd (curve F–G);

a final contraction.

For the Nd96Cd4 sample the eutectoid reaction at 500 °C was indicated in the dilatometric curve during heating (A in Fig. 6b). DTA measurements of the latter sample did not show an effect which could be ascribed to the eutectoid reaction, probably because of the low sensitivity. A rather implausible contraction can be observed after point A in the dilatometric curve which was probably caused by a contraction of the sample during the eutectoid reaction which caused it to slip out of the specimen holder inside the dilatometer.

Fig. 7 shows the Nd-rich part of the Nd–Cd phase diagram with the effects recorded DTA and dilatometry (with a heating rate of 5 K min−1). Obviously, these experimental data agree very well.

Fig. 7

Nd-rich part of the Nd–Cd phase diagram together with effects from DTA and dilatometry, recorded with a heating rate of 5 K min−1.

Summary

The Nd–Cd phase diagram was constructed using a combination of DTA, SEM, XRD and dilatometry data (Fig. 1). Cd solubilities in both α-Nd and β-Nd were determined and are listed in Table 1. Seven intermetallic compounds, described previously in literature, were confirmed and their phase boundaries were estimated by a combination of all available results. A congruent formation of NdCd, NdCd2 and Nd11Cd45 was found whereas NdCd3, Nd13Cd58, NdCd6 and NdCd11 decompose incongruently. The temperature values of the peritectic formation of NdCd11 and of the eutectic between NdCd11 and Cd were comparable with those given in Ref. [20], other invariant reaction temperatures had to be modified with respect to the literature values. Due to DTA results a polymorphic transition of NdCd2 is proposed although the high temperature form β-NdCd could not be quenched. X-ray diffraction was used to confirm the crystal structure of all intermetallic compounds.

A dilatometric study was performed for Nd–Cd samples up to 4 at.% Cd. The results fit well with DTA data. α-Nd seems to have a higher thermal expansion than β-Nd.