Synthesis of Intermetallic (Mg1−x,Alx)2Ca by Combinatorial Sputtering

The synthesis–composition–structure relationship in the Mg–Ca–Al system is studied using combinatorial magnetron sputtering. With increasing deposition temperature, a drastic decrease in Mg concentration is obtained. This behavior can be understood based on density functional theory calculations yielding a desorption energy of 1.9 eV/atom for Mg from a hexagonal Mg nanocluster which is far below the desorption energy of Mg from a Mg2Ca nanocluster (3.4 eV/atom) implying desorption of excess Mg during thin film growth at elevated temperatures. Correlative structural and chemical analysis of binary Mg–Ca thin films suggests the formation of hexagonal Mg2Ca (C14 Laves phase) in a wide Mg/Ca range from 1.7 to 2.2, expanding the to date reported stoichiometry range. Pronounced thermally-induced desorption of Mg is utilized to synthesize stoichiometric (Mg1−x,Alx)2Ca thin films by additional co-sputtering of elemental Al, exhibiting a higher desorption energy (6.7 eV/atom) compared to Mg (3.4 eV/atom) from Mg2Ca, which governs its preferred incorporation during synthesis. X-ray diffraction investigations along the chemical gradient suggest the formation of intermetallic C14 (Mg1–x,Alx)2Ca with a critical aluminum concentration of up to 23 at.%. The introduced synthesis strategy, based on the thermally-induced desorption of weakly bonded species, and the preferential incorporation of strongly bonded species, may also be useful for solubility studies of other phases within this ternary system as well as for other intermetallics with weakly bonded alloying constituents.


Introduction
Hexagonal close-packed magnesium (Mg) combines promising properties for structural applications considering that it exhibits one of the lowest densities among metals (1.7 g/cm 3 ) as well as a high specific strength, high abundance, and good castability [1,2]. The main alloying element for commercial Mg alloys is aluminum (Al), promoting corrosion resistance [3]. Nevertheless, most Mg-Al alloys undergo a rapid degradation of creep resistance at elevated temperatures restricting their implementation in industry to non-critical parts [1,4,5]. However, the addition of calcium (Ca) can enhance the creep resistance, which is commonly reasoned by the precipitation of intermetallic (Mg,Al) 2 Ca Laves phases [4][5][6][7][8].
Within the ternary Mg-Al-Ca system, the cubic (C15) Al 2 Ca and the hexagonal (C14) Mg 2 Ca phases are reported to be the thermodynamically stable binary Laves phases with an extensive solubility of the third constituent, Mg in Al 2 Ca and Al in Mg 2 Ca [9]. Recent experiments and first-principle calculations also suggest the existence of a dihexagonal C36 (Mg,Al) 2 Ca phase [9][10][11][12].

Materials and Methods
Thin film depositions were carried out by magnetron sputtering in a laboratory-scale high vacuum chamber evacuated to a base pressure below 1 × 10 −4 Pa. All depositions were conducted at a constant argon partial pressure of 0.5 Pa. The utilized Si (100) substrates were positioned at a target-to-substrate distance of 10 cm. The set temperature, corresponding to the substrate temperature prior to the ignition of the plasma, was varied between room temperature (no intentional heating) and 300 • C. A circular Mg-Ca target (50 mm in diameter) was produced from a purchased Mg-Mg 2 Ca composite material with a composition of Mg83-Ca17 at.% determined by energy dispersive X-ray analysis (EDX). The target was operated in pulsed direct-current mode with an ENI RPG-100E generator at a frequency of 250 kHz and duty cycle of 60%. The target was powered with 60 and 200 W for the deposition of binary Mg-Ca and ternary Mg-Ca-Al samples, respectively. The Al target (99.99% purity; 50 mm in diameter) power was varied from 20 to 100 W. While the composite target was facing the substrate, the Al target was mounted at an inclination angle of 45 • relative to the substrate normal resulting in a compositional gradient (see below).
Chemical composition analysis of the combinatorial Mg-Ca-Al films was conducted by EDX utilizing a JEOL JSM-6480 scanning electron microscope (JEOL Ltd., Tokyo, Japan) equipped with an EDAX Genesis 2000 analyzer (EDAX Inc., Mahwah, NJ, USA). An acceleration voltage of 8 kV was used. Spatially-resolved structural analysis was conducted by X-ray diffraction (XRD) using a Bruker AXS D8 Discover (Bruker Corporation, Billerica, MA, USA) General Area Detector Diffraction System (GADDS) with a pinhole-collimated primary beam of 0.5 mm. The incidence angle ω was fixed to 15 • whereas the 2θ angle was increased from 15 • to 75 • during the measurements. The voltage and current of the Cu Kα radiation sources were set to 40 kV and 40 mA, respectively. For the structural identification of hexagonal Mg, the JCPDS card no. 35-0821 was used.
To estimate desorption energies of Mg, Ca, and Al adatoms from Mg and Mg 2 Ca nanoclusters, mimicking fundamental surface processes during the growth of Mg-Ca thin films with Al additions, density functional theory (DFT) [18] at 0 K was employed. A linear combination of localized quasiatomic orbitals [19] and the generalized gradient approximation was used [20] within the OpenMX software package [21]. The basis functions were defined employing a confinement scheme [19,22] and described as: Mg7.0-s3p3d2, Ca9.0-s4p3d2, and Al7.0-s3p3d2 (the leading symbol denotes the chemical element, followed by the cut-off radius in the Bohr radii, and the last symbol designates the primitive orbitals). An energy cut-off of 150 Ry was chosen to reach a total energy convergence of 10 −6 Ry. No periodic boundary conditions were applied. The desorption energy, defined herein as positive, was calculated by evaluating the total energy difference between the nanocluster + adatom (an adatom at a bulk equilibrium position) and the isolated nanocluster + isolated adatom. A Mg nanocluster was constructed by adopting an isostructural (hexagonal close-packed) Ti nanocluster containing 13 atoms [23]. A Mg terminated Mg 2 Ca nanocluster was formed based on an isostructural hexagonal C14 Fe 2 W nanocluster comprised of 125 atoms [24].

Binary Mg-Ca
First, the synthesis-composition-structure relationship is studied for binary Mg-Ca. Depositions were conducted at varying set temperatures between room temperature and 300 • C. The Mg-Ca thin films are characterized regarding their chemical composition ( Figure 1a) and crystal structure ( Figure 1b).  [19,22] and described as: Mg7.0-s3p3d2, Ca9.0-s4p3d2, and Al7.0-s3p3d2 (the leading symbol denotes the chemical element, followed by the cut-off radius in the Bohr radii, and the last symbol designates the primitive orbitals). An energy cut-off of 150 Ry was chosen to reach a total energy convergence of 10 −6 Ry. No periodic boundary conditions were applied. The desorption energy, defined herein as positive, was calculated by evaluating the total energy difference between the nanocluster + adatom (an adatom at a bulk equilibrium position) and the isolated nanocluster + isolated adatom. A Mg nanocluster was constructed by adopting an isostructural (hexagonal close-packed) Ti nanocluster containing 13 atoms [23]. A Mg terminated Mg2Ca nanocluster was formed based on an isostructural hexagonal C14 Fe2W nanocluster comprised of 125 atoms [24].

Binary Mg-Ca
First, the synthesis-composition-structure relationship is studied for binary Mg-Ca. Depositions were conducted at varying set temperatures between room temperature and 300 °C. The Mg-Ca thin films are characterized regarding their chemical composition ( Figure 1a  For the samples deposited without intentional heating and 100 °C, the measured Mg/Ca ratios match the composition of the utilized composite Mg-Mg2Ca sputtering target. With a further increase in set temperature, the Mg/Ca ratio exhibits a drastic decrease with no Mg detectable in the films synthesized at temperatures ≥250 °C. Consequently, within the temperature range from 150 to 250 °C the chemical composition is extremely sensitive to minute temperature changes implying thermallyinduced desorption of Mg during thin film growth as the species with the higher vapor pressure compared to Ca [25]. Furthermore, the weight of the thin films was determined by measuring the weight of the substrate prior to and after the deposition. The functional dependence of the thin film weight on temperature agrees well with the measured trend in Mg/Ca ratio supporting the notion of Mg desorption. Due to partial desorption of Mg, the sample synthesized at 200 °C exhibits the stoichiometry of intermetallic Mg2Ca. XRD analysis revealed an amorphous or nanocrystalline structure of the sample synthesized at room temperature. Upon a temperature increase to 100 and 150 °C, representative peaks for hexagonal Mg at 37° and 63° as well as for Mg2Ca [26] at 29°, 31°, 34° For the samples deposited without intentional heating and 100 • C, the measured Mg/Ca ratios match the composition of the utilized composite Mg-Mg 2 Ca sputtering target. With a further increase in set temperature, the Mg/Ca ratio exhibits a drastic decrease with no Mg detectable in the films synthesized at temperatures ≥250 • C. Consequently, within the temperature range from 150 to 250 • C the chemical composition is extremely sensitive to minute temperature changes implying thermally-induced desorption of Mg during thin film growth as the species with the higher vapor pressure compared to Ca [25]. Furthermore, the weight of the thin films was determined by measuring the weight of the substrate prior to and after the deposition. The functional dependence of the thin film weight on temperature agrees well with the measured trend in Mg/Ca ratio supporting the notion of Mg desorption. Due to partial desorption of Mg, the sample synthesized at 200 • C exhibits the stoichiometry of intermetallic Mg 2 Ca. XRD analysis revealed an amorphous or nanocrystalline structure of the sample synthesized at room temperature. Upon a temperature increase to 100 and  2 Ca phase mixture, being in agreement with chemical composition data. It may be speculated that the peak shift of 0.5 • for the Mg 2 Ca diffraction peak at 29 • towards higher diffraction angles [26] originates from an off-stoichiometric Laves phase composition. Since the sample is enriched in Mg compared to stoichiometric Mg 2 Ca, some Ca atoms may be replaced by smaller Mg atoms, hence, leading to a decrease in lattice spacing as observed by the shifts to higher 2θ values. Off-stoichiometric formation of C14 Mg 2 Ca will further be discussed below. For the film deposited at 200 • C, no characteristic Mg peaks are obtained any longer, but all acquired diffraction peaks are attributed to intermetallic Mg 2 Ca in accordance with the measured composition. Hence, XRD phase pure C14 Mg 2 Ca is obtained at 200 • C, despite the usage of a Mg enriched target, due to desorption of excess Mg. However, at set temperatures above 200 • C, no Mg remains in the growing film as indicated by both, the chemical and structural analysis.
As a next step, the set temperature range between 150 and 250 • C, identified as critical for the resulting thin film composition and, hence, structure evolution, was studied in more detail. For this purpose, the set temperature was varied in steps of 10 • C between the individual depositions. The results of the experimental and theoretical study are illustrated in Figure 2. and 35.5° are obtained indicating the presence of a Mg-Mg2Ca phase mixture, being in agreement with chemical composition data. It may be speculated that the peak shift of 0.5° for the Mg2Ca diffraction peak at 29° towards higher diffraction angles [26] originates from an off-stoichiometric Laves phase composition. Since the sample is enriched in Mg compared to stoichiometric Mg2Ca, some Ca atoms may be replaced by smaller Mg atoms, hence, leading to a decrease in lattice spacing as observed by the shifts to higher 2θ values. Off-stoichiometric formation of C14 Mg2Ca will further be discussed below. For the film deposited at 200 °C, no characteristic Mg peaks are obtained any longer, but all acquired diffraction peaks are attributed to intermetallic Mg2Ca in accordance with the measured composition. Hence, XRD phase pure C14 Mg2Ca is obtained at 200 °C, despite the usage of a Mg enriched target, due to desorption of excess Mg. However, at set temperatures above 200 °C, no Mg remains in the growing film as indicated by both, the chemical and structural analysis. As a next step, the set temperature range between 150 and 250 °C, identified as critical for the resulting thin film composition and, hence, structure evolution, was studied in more detail. For this purpose, the set temperature was varied in steps of 10 °C between the individual depositions. The results of the experimental and theoretical study are illustrated in Figure 2. In the investigated temperature range in Figure 2, the Mg/Ca ratio shows a trend comparable to a tangent function, separating two temperature regions in which a drastic decrease in Mg/Ca is obtained. XRD revealed a phase mixture of Mg and Mg2Ca for the thin films synthesized at 150, 160, and 170 °C. Between 180 and 230 °C, the Mg/Ca ratio gradually but slightly decreases from 2.2 to 1.7, which corresponds to Mg concentrations of 68.5 and 63 at.%, respectively. Hence, the results suggest off-stoichiometric formation of intermetallic Mg2Ca in the given composition range, extending the range reported by Suzuki et al., who found a solubility of excess Mg in Mg2Ca of 4.5 at.% using bulk samples [13]. In contrast, Kevorkov et al. [14] reported Mg2Ca to be a line compound based on a diffusion couple study. At 240 °C, all sputtered Mg is desorbed from the surface resulting in the formation of fcc Ca.
The obtained non-linear temperature dependency of the Mg/Ca ratio, following the trend of a tangent function, clearly cannot be explained based on Mg vapor pressure considerations alone. To mimic surface processes active during the growth of Mg-Ca thin films, desorption energies of a Mg adatom on hexagonal close-packed Mg and a hexagonal C14 Mg2Ca nanocluster were calculated using DFT, yielding 1.9 and 3.4 eV/atom, respectively. These values are lower than the corresponding predicted desorption energies for Ca of 2.4 and 4.2 eV/atom, respectively. For comparison, required In the investigated temperature range in Figure 2, the Mg/Ca ratio shows a trend comparable to a tangent function, separating two temperature regions in which a drastic decrease in Mg/Ca is obtained. XRD revealed a phase mixture of Mg and Mg 2 Ca for the thin films synthesized at 150, 160, and 170 • C. Between 180 and 230 • C, the Mg/Ca ratio gradually but slightly decreases from 2.2 to 1.7, which corresponds to Mg concentrations of 68.5 and 63 at.%, respectively. Hence, the results suggest off-stoichiometric formation of intermetallic Mg 2 Ca in the given composition range, extending the range reported by Suzuki et al., who found a solubility of excess Mg in Mg 2 Ca of 4.5 at.% using bulk samples [13]. In contrast, Kevorkov et al. [14] reported Mg 2 Ca to be a line compound based on a diffusion couple study. At 240 • C, all sputtered Mg is desorbed from the surface resulting in the formation of fcc Ca.
The obtained non-linear temperature dependency of the Mg/Ca ratio, following the trend of a tangent function, clearly cannot be explained based on Mg vapor pressure considerations alone. To mimic surface processes active during the growth of Mg-Ca thin films, desorption energies of a Mg adatom on hexagonal close-packed Mg and a hexagonal C14 Mg 2 Ca nanocluster were calculated using DFT, yielding 1.9 and 3.4 eV/atom, respectively. These values are lower than the corresponding predicted desorption energies for Ca of 2.4 and 4.2 eV/atom, respectively. For comparison, required desorption energies for K and Na from Al (001) and Al (111) are reported to lie in the range between 1.1 and 1.8 eV/atom [27]. The energetically favored Mg desorption compared to the desorption of Ca for both considered configurations is in agreement with the experimentally observed Mg loss. Furthermore, Mg desorption from a Mg nanocluster is energetically favored compared to the Mg desorption from a Mg 2 Ca nanocluster, and is, hence, suggested as the governing mechanism for set temperatures of up to~170 • C. In contrast, the adatom Mg on Mg 2 Ca, due to the stronger bond formation between Mg and Ca, is not desorbing and readily incorporated into the growing film. As a consequence, in the temperature range of~180 to 230 • C, the formation of intermetallic Mg 2 Ca is observed, in which all weakly bonded excess Mg is desorbed. At a set temperature of 240 • C, the formation of fcc Ca is observed indicating the desorption of all sputtered Mg from the thin film surface. Rapid degradation of the crystal quality of fcc Ca upon exposure to atmosphere, due to its high reactivity with oxygen, is observed (see absence of crystalline peaks in Figure 1b at a set temperature of 250 • C).

Ternary Mg-Ca-Al
Based on the data for sputtered Mg-Ca thin films, desorption of excess Mg has been identified as a beneficial mechanism for the synthesis of stoichiometric intermetallic Mg 2 Ca. To study the synthesis-composition-structure relationship along the quasi-binary Mg 2 Ca-Al 2 Ca, stoichiometric (Mg,Al) 2 Ca samples are required. Here, we seek to exploit the above identified desorption mechanism in Mg 2 Ca to grow stoichiometric (Mg,Al) 2 Ca thin films. For this purpose, the desorption energy of Al from a Mg 2 Ca nanocluster is predicted using DFT yielding 6.7 eV/atom, which is compared to the desorption energy of Mg of 3.4 eV/atom. Hence, the energetic barrier for Al desorption is considerably higher than that for Mg, indicating the formation of stronger bonds between an Al adatom and underlying intermetallic Mg 2 Ca compared to a Mg adatom. In comparison, the desorption energy of Al on a Mg 2 Ca nanocluster lies in the range of O desorption from fcc Al(111), which has been reported to require 7.3 eV/atom [28]. Consequently, our calculations suggest that Al is preferably incorporated into the intermetallic (Mg,Al) 2 Ca phases. By choosing a set temperature above the desorption temperature of excess Mg, e.g., 200 • C (see Section 3.1), we seek to exploit the above identified preferential desorption of excess Mg to grow stoichiometric (Mg,Al) 2 Ca thin films along the Mg 2 Ca-Al 2 Ca gradient. In order to critically appraise this hypothesis, Mg-Al-Ca thin films were deposited by co-sputtering of Al with varying DC target power values as well as set temperatures. The deposition setup and the resulting gradient in composition is schematically illustrated in Figure 3a. The obtained thin film composition, measured along the gradient for the various samples by EDX, is depicted in Figure 3b.
At room temperature (approximately 20 • C), a chemical gradient for all sputtered species, i.e., Al, Mg, and Ca, was obtained (see Figure 3b), which can be expected based on the employed deposition setup (see Figure 3a). Furthermore, a constant Mg/Ca ratio was measured along the Al gradient, proving that Mg is not desorbing, as anticipated for such a low temperature. Upon an increase in set temperature to 200 • C, the gradient in Ca vanished and the Mg concentration was significantly reduced. As a result, the measured thin film compositions lie on the quasi-binary of Mg 2 Ca and Al 2 Ca with variations in the measured Ca concentration of less than 0.5 at.%. With increasing Al target power, the measured Al concentration increased solely at the expense of Mg resulting in thin film stoichiometries of (Mg 1−x ,Al x ) 2 Ca with x ranging from 0 to 0.9. These results verify the hypothesis, derived based on the DFT predictions, that the sputtered Al replaces the Mg in (Mg,Al) 2 Ca, subsequently triggering its desorption as observed already for binary Mg-Ca. Thus, a synthesis strategy for stoichiometric Mg-Ca-Al thin films along the quasi-binary Mg 2 Ca-Al 2 Ca is proposed and validated, enabling the efficient study of the composition-induced phase formation. The results are presented in Figure 4. exploit the above identified preferential desorption of excess Mg to grow stoichiometric (Mg,Al)2Ca thin films along the Mg2Ca-Al2Ca gradient. In order to critically appraise this hypothesis, Mg-Al-Ca thin films were deposited by co-sputtering of Al with varying DC target power values as well as set temperatures. The deposition setup and the resulting gradient in composition is schematically illustrated in Figure 3a. The obtained thin film composition, measured along the gradient for the various samples by EDX, is depicted in Figure 3b.  At room temperature (approximately 20 °C), a chemical gradient for all sputtered species, i.e., Al, Mg, and Ca, was obtained (see Figure 3b), which can be expected based on the employed deposition setup (see Figure 3a). Furthermore, a constant Mg/Ca ratio was measured along the Al gradient, proving that Mg is not desorbing, as anticipated for such a low temperature. Upon an increase in set temperature to 200 °C, the gradient in Ca vanished and the Mg concentration was significantly reduced. As a result, the measured thin film compositions lie on the quasi-binary of Mg2Ca and Al2Ca with variations in the measured Ca concentration of less than 0.5 at.%. With increasing Al target power, the measured Al concentration increased solely at the expense of Mg resulting in thin film stoichiometries of (Mg1−x,Alx)2Ca with x ranging from 0 to 0.9. These results verify the hypothesis, derived based on the DFT predictions, that the sputtered Al replaces the Mg in (Mg,Al)2Ca, subsequently triggering its desorption as observed already for binary Mg-Ca. Thus, a synthesis strategy for stoichiometric Mg-Ca-Al thin films along the quasi-binary Mg2Ca-Al2Ca is proposed and validated, enabling the efficient study of the composition-induced phase formation. The results are presented in Figure 4.  [14]; Amerioun et al. [11]). XRD analysis revealed that all obtained peaks can be attributed to the hexagonal C14 structure of Mg2Ca [26] for Al concentrations between 5.6 and 23.3 at.%. With an increase in Al concentration, a continuous peak shift towards larger 2θ angles is obtained, indicating Al incorporation into the C14 structure at least up to 23.3 at.%. These data are in agreement with bulk studies, reporting an Al solubility of 22 [13] and 25 at.% [14], respectively, supporting that Mg2Ca exhibits a significantly higher critical Al solubility than 16 at.%, as claimed by Amerioun et al. [11]. With increasing Al concentration from 5.6 to 23.3 at.%, an increase in full width at half maximum by a factor of 2.7 is obtained, evaluated based on the (103) peak. The peak broadening at even higher Al concentrations (not shown), eventually indicating a decreased crystal quality or competitive phase formation, impedes a meaningful discussion of the XRD data at higher alloying concentrations.
By employing Bragg´s law [29], the lattice parameters a and c can be calculated based on the XRD data. The results are depicted in Figure 4b. The lattice parameters of binary Mg2Ca synthesized at 200 °C are also added. With increasing Al concentrations solved in hexagonal C14 Mg2Ca, a continuous decrease in lattice parameters a and c is obtained, which can be reasoned by the smaller XRD analysis revealed that all obtained peaks can be attributed to the hexagonal C14 structure of Mg 2 Ca [26] for Al concentrations between 5.6 and 23.3 at.%. With an increase in Al concentration, a continuous peak shift towards larger 2θ angles is obtained, indicating Al incorporation into the C14 structure at least up to 23.3 at.%. These data are in agreement with bulk studies, reporting an Al solubility of 22 [13] and 25 at.% [14], respectively, supporting that Mg 2 Ca exhibits a significantly higher critical Al solubility than 16 at.%, as claimed by Amerioun et al. [11]. With increasing Al concentration from 5.6 to 23.3 at.%, an increase in full width at half maximum by a factor of 2.7 is obtained, evaluated based on the (103) peak. The peak broadening at even higher Al concentrations (not shown), eventually indicating a decreased crystal quality or competitive phase formation, impedes a meaningful discussion of the XRD data at higher alloying concentrations.
By employing Bragg's law [29], the lattice parameters a and c can be calculated based on the XRD data. The results are depicted in Figure 4b. The lattice parameters of binary Mg 2 Ca synthesized at 200 • C are also added. With increasing Al concentrations solved in hexagonal C14 Mg 2 Ca, a continuous decrease in lattice parameters a and c is obtained, which can be reasoned by the smaller atomic radius of Al (1.25 Å) compared to Mg (1.4 Å) [30]. Calculation of the hypothetical C14 Al 2 Ca phase yields lattice parameters a and c of 5.67 and 9.23 Å [31], respectively, which are consistent with the composition induced change in lattice parameters measured here. The obtained trend is also consistent with bulk studies, showing a good quantitative agreement with Kevorkov et al. [14], while the decrease in lattice parameters reported by Amerioun et al. [11] appears to be overestimated. It should be noted that compositional gaps in the composition-structure relationship known for the bulk samples are now filled employing the combinatorial thin film methodology. Considering the high reactivity of Ca, hampering experimental investigations, the pioneered usage of less-reactive intermetallic Mg 2 Ca as a source for Ca, presented within this work, serves as a sophisticated approach to systematically study the Mg-Ca-Al system.

Conclusions
Combinatorial magnetron sputtering from a composite Mg 2 Ca-Mg (Mg83-Ca17 at.%) and an elemental Al target was utilized to synthesize Mg-Ca-Al thin films. The obtained Mg concentration was found to be extremely sensitive to minute changes in set temperature (substrate temperature prior to plasma ignition). To mimic processes active on the thin film surface during synthesis, desorption energies of the individual species from a hexagonal close-packed Mg and a hexagonal C14 Mg 2 Ca nanocluster were calculated using DFT. The calculations suggest a desorption energy of 1.9 eV/atom for Mg from a hexagonal Mg nanocluster which is significantly lower than the desorption energy of Mg from a Mg 2 Ca nanocluster (3.4 eV/atom), implying temperature-induced desorption of excess Mg during thin film growth at elevated temperatures. Consequently, XRD phase pure Mg 2 Ca was synthesized at set temperatures between 180 and 230 • C due to desorption of excessively sputtered Mg that cannot be incorporated in Mg 2 Ca. Correlative structural and chemical analysis suggests the off-stoichiometric formation of Mg 2 Ca in a Mg/Ca ratio range from 1.7 to 2.2. Above 240 • C, the formation of fcc Ca indicates desorption of all sputtered Mg from the thin film surface. Furthermore, the desorption energy of Al from Mg 2 Ca was calculated to be 6.7 eV/atom, which is almost twice the desorption energy of Mg from a Mg 2 Ca nanocluster, implying a preferred incorporation of Al instead of Mg in (Mg,Al) 2 Ca during sputtering. Consequently, this mechanism of excess Mg evaporation at elevated synthesis temperatures was exploited experimentally by co-sputtering of Al, resulting in the formation of stoichiometric chemically-graded (Mg 1−x ,Al x ) 2 Ca with x varying from 0 to 0.9. Systematic investigation of the composition-structure relationship along the chemical gradient suggests the formation of hexagonal C14 Mg 2 Ca with a significantly higher critical aluminum concentration as reported by Amerioun et al. [11], in agreement with Suzuki et al. [13] and Kevorkov et al. [14]. The introduced synthesis strategy, based on the thermally-induced desorption of weakly bonded species, and the preferential incorporation of strongly bonded species, enabled the systematic investigation of the critical Al solubility of the ternary (Mg 1−x ,Al x ) 2 Ca intermetallic phase. This approach may also be useful for solubility studies of other phases within this ternary system as well as for other intermetallics exhibiting weakly and stongly bonded alloying constituents.