Data on atomic structures of precipitates in an Al-Mg-Cu alloy studied by high resolution transmission electron microscopy and first-principles calculations

The dataset refers to the research article “Precipitation processes and structural evolutions of various GPB zones and two types of S phases in a cold-rolled Al-Mg-Cu alloy” [1]. Transmission electron microscopy (TEM) and density functional theory (DFT) were used to investigate precipitates in an Al-Cu-Mg alloy aged at 443 K for various times. High-angle annular dark-field scanning TEM (HAADF-STEM) images in <100> Al orientations were analyzed. Characteristic contrast and symmetries of columns [2] yielded atoms and positions, used to build precipitate models which could be refined and compared with solid solution reference energies. A calculation cell is an Al supercell compatible with symmetry and morphology of a precipitate, which is fully or partly surrounded by Al, allowing periodicity continuation via neighbor cells. The given crystallographic data include two S-phase variants and Guinier–Preston–Bagaryatsky (GPB) zones, of which the “GPBX” is new.


a b s t r a c t
The dataset refers to the research article "Precipitation processes and structural evolutions of various GPB zones and two types of S phases in a cold-rolled Al-Mg-Cu alloy" [1] . Transmission electron microscopy (TEM) and density functional theory (DFT) were used to investigate precipitates in an Al-Cu-Mg alloy aged at 443 K for various times. Highangle annular dark-field scanning TEM (HAADF-STEM) images in < 100 > Al orientations were analyzed. Characteristic contrast and symmetries of columns [2] yielded atoms and positions, used to build precipitate models which could be refined and compared with solid solution reference energies. A calculation cell is an Al supercell compatible with symmetry and morphology of a precipitate, which is fully or partly surrounded by Al, allowing periodicity continuation via neighbor cells. The given crystallographic data in-  Table   Subject Materials Science Specific subject area Precipitates in Al-Mg-Cu alloys Type of data 1. Image (HAADF-STEM images) 2. Table (Crystal structure data of precipitate models and refinements) How data were acquired 1. In a transmission electron microscope (double Cs corrected JEOL ARM 200F). 2. Models manually extracted from HAADF images. Images were contrast enhanced using Gatan Microscopy Suite (GMS) and the freely available computer software package ImageJ. Atoms were identified using rules of columns [2] . Interatomic distances and symmetry operations were checked, and space group determined using VESTA [3] . The models were input and refined using the Vienna ab-initio simulation package (VASP). Data format Raw Analyzed Filtered Parameters for data collection 1. Precipitates were imaged using a HAADF in STEM mode, in < 100 > Al orientations of regions of thickness typically less than 5 nm. 2. The models were manually extracted from the images, calculation cells tried out, input in VESTA, where after suitable cells were determined, and initial atomic coordinates calibrated using the assumed Al-Al inter-column distance 2.025 Å . The calculations were performed at zero Kelvin using the projector augmented wave method within the PBE (Perdew-Burke-Ernzerhof) generalized gradient approximation [4] . The plane-wave energy cut-off was 400 eV and a Monkhorst-Pack gamma-centered k-point mesh was used, with maximal k-point distances of 0.18 Å −1 in each direction [5] . The electronic accuracy for self-consistent loops was set at 10 −6 eV and the atomic positions were relaxed to a maximal force of 0.001 eV/ Å between atoms using first-order Methfessel-Paxton for smearing of partial occupation and a smearing factor (SIGMA) of 0.

Value of the Data
• The atomic resolution HAADF-STEM data reported here clearly show the microstructures of Al-Mg-Cu alloys under different conditions, especially the atomic structure of various GPB zones and S phases formed during the aging treatment. The atomic models of precipitates constructed in this article give us a more intuitive impression of the precipitated phase structures. DFT calculation helps to understand the precipitation process. • Researchers who investigate the precipitation process or precipitates structures in Al-Mg-Cu alloys can benefit from the data. • The data can be used as a reference for other researchers to obtain specific precipitates and to optimize the precipitation heat treatment conditions. Fig. 1 shows the HAADF-STEM images of the as-quenched (A.Q.) and as-rolled (A.R.) samples before aging. There is no precipitate found in the Al matrix ( Fig. 1 (a) and (d)) or near the defects (the dislocation loop in Fig. 1 (b) and the grain boundary in Fig. 1 (c)). Fig. 2 shows two types of S phases [1] formed along dislocations in the A.Q. sample aged at 443 K for 8 h. Fig. 3 and 4 are HAADF-STEM images observed in < 100 > Al orientation in the 7% cold-rolled Al-3Mg-1Cu alloy aged at 443 K for 20 min. Fig. 3 shows several GPB zones and the GPBX zone as well as the thinnest S-I phase on a dislocation line. The marked area containing the GPBX zone corresponds to Fig. 2 (c) of reference [1] . Several GPBX zones and the thinnest S-I phase as well as a forming S-II phase are found to form along a dislocation line as shown in Fig. 4 . The selected area is magnified and atomic overlaid to be Fig. 7 of reference [1] .     Table 1 shows the numerical values from DFT calculations of the precipitate structure models presented in Fig. 4 of reference [1] , including formation enthalpy per solute atom, pressure and composition. The refined fractional coordinates in the DFT calculated atomic model containing various GPB zones and two types of S phases are listed in Tables 2-13 . In all of the atomic  models, the precipitates are embedded in the Al matrix, so the P1 space group is used to give  refined coordinates.  Tables 2 and 3 represent the models of the GPB 1 zone with Cu or Al interstitial, respectively. Tables 4 and 5 show models containing a single GPB 2 zone or a pair of GPB 2 zones, respectively. Table 6 exhibits the model of the newly observed GPBX zone. Tables 7 and 8 show models of two individual GPBX zones orientated along < 114 > Al or < 113 > Al directions, respectively. The illustration can be found in reference [1] . Tables 9 , 10 and 11 display the relaxed models of several GPBX zones listed in Tables 6 , 7 and 8 , respectively. Tables 12 and 13 show the models of S-I and S-II phases, respectively. Note that the corresponding raw atomic models of all precipitates listed in Tables 2 to 13 can be found in the supplementary materials as cif files.    Table 3 The refined fractional coordinates in the atomic model for the GPB 1 (Al interstitial) zone.        Table 9 The refined fractional coordinates in the atomic model for the GPBX_relax zone.

Materials
The investigated Al-Mg-Cu alloy (3.04% Mg, 1.01% Cu, 0.01% Si, 0.01% Fe, and Al of the remaining amount, all by mass) was provided by Furukawa sky (now, UACJ Corporation). In the factory, materials were mold cast into 400 × 170 × 40 mm bulk materials and homogenized at 773 K for 10 h. After the facing work, the billets were rolled into 3 mm thick sheets at 683 K. Finally, the sheets were cold rolled to a thickness of 1.2 mm.
The received materials were cut to small sheet-shaped samples of dimension 10 × 10 × 1.2 (in mm), and then solution heat treated (SHT) in a salt bath at 793 K for 1 h, followed by quenching in iced water (273 K). After SHT, cold rolling (CR) was performed at room temperature (approximately 293 K) with a reduction rate of 7%. The change in thickness of the sheet was monitored to determine the reduction rate. Isothermal artificial aging was performed on the A .Q. and A .R. samples in an oil bath at 443 K for 20 min and 8 h, respectively. The as-quenched and as-rolled conditions are abbreviated as "A.Q." and "A.R." respectively.

Transmission electron microscopy
For the preparation of TEM specimens, 1.2 mm thick sheet samples were first ground down to around 100 μm thickness, and then from 3 mm diameter discs were punched out of ground thin sheets. Subsequently, a Struers TenuPol-5 machine was used to electropolishing the discs until perforation. Two liters of electrolyte was prepared by mixing the 667 ml nitric acid (60.0% concentration) and 1333 ml methanol (99.0% concentration), and the temperature was kept between 243 K and 253 K during electropolishing. To reduce the amount of contamination, prior to the HAADF-STEM observations, all specimens were plasma cleaned for 3 min in a Fischione 1020 Plasma Cleaner. All the high-resolution HAADF-STEM images in this article were taken in a < 100 > Al orientation, as precipitates extending along this direction. The instrument was a double Cs corrected JEOL ARM 200F operated at 200 kV. The convergence semi-angle was set to 28 mrad and the inner collection angle of the HAADF detector was 48 mrad.
Some of the HAADF-STEM images were filtered using a circular bandpass mask applied on the respective fast Fourier transform (FFT) to further improve the clarity, and an inverse FFT (IFFT) was performed on the masked area. This operation can cut all spatial frequencies that correspond to features in the real space smaller than 0.15 nm. The operation was performed in the software "GMS 3".

First-principles calculations
The first principles calculations were carried out with density functional theory as implemented in the Vienna ab initio simulation package (VASP) [ 6 , 7 ]. All formation enthalpies were calculated at zero Kelvin using the projector augmented wave method within the PBE (Perdew-Burke-Ernzerhof) generalized gradient approximation [4] with a plane-wave energy cut-off of 400 eV. A Monkhorst-Pack gamma-centered k-point mesh was used, with maximal k-point distances of 0.18 Å −1 in each direction [5] . Partial occupancies were smeared using the first-order Methfessel-Paxton method with a smearing factor (SIGMA) of 0.2. For final energies, the tetrahedron method with Blöchl correction for the smearing [8] . The convergence criteria for the electronic convergence in the self-consistent cycles was 10 −6 eV and the atomic positions were relaxed to a maximal atomic force of 1 meV/ Å .
The zero Kelvin formation enthalpies were calculated as described by Marioara et al. [9] using a fixed aluminum lattice parameter of 4.0400 Å , corresponding to the lattice parameter relaxed with VASP using the above parameters. Not relaxing the supercell size makes it easier to compare different calculations but also overestimates the strain contribution to the formation enthalpy (corresponding to the surrounding Al being infinitely hard). A lower limit to the strain contribution was found for the structures with the largest internal pressure as listed in Table 1 by fully relaxing the supercell size (corresponding to the surrounding Al being infinitely soft). To reduce systematic errors arising from the k-point meshing of different supercell sizes, separate reference calculations for a single solute atom in the Al lattice were performed for each supercell size.