Experimental and DFT characterization of η′ nano-phase and its interfaces in AlZnMgCu alloys

Accepted Manuscript Experimental and DFT characterization of η′ nano-phase and its interfaces in Al-ZnMg-Cu alloys Fuhua Cao, Jingxu Zheng, Yong Jiang, Bin Chen, Yiren Wang, Tao Hu PII:

S1359-6454(18)30845-0

DOI:

https://doi.org/10.1016/j.actamat.2018.10.045

Reference:

AM 14923

To appear in:

Acta Materialia

Received Date: 23 August 2018 Revised Date:

18 October 2018

Accepted Date: 22 October 2018

Please cite this article as: F. Cao, J. Zheng, Y. Jiang, B. Chen, Y. Wang, T. Hu, Experimental and DFT characterization of η′ nano-phase and its interfaces in Al-Zn-Mg-Cu alloys, Acta Materialia, https:// doi.org/10.1016/j.actamat.2018.10.045. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

Graphical abstract:

ACCEPTED MANUSCRIPT

Experimental and DFT characterization of η′ nano-phase and its interfaces in Al-Zn-Mg-Cu alloys

a

RI PT

Fuhua Caoa, Jingxu Zhengb,c, Yong Jianga,d∗, Bin Chenb∗∗, Yiren Wanga, Tao Hua,d

Key Laboratory for Nonferrous Materials (MOE), School of Materials Science and Engineering, Central South University, Changsha 410083, China

School of Materials Science and Engineering, Shanghai Jiao Tong University, Shanghai 200240, China

c

Department of Materials Science and Engineering, Cornell University, Ithaca, NY 14853, USA

d

State Key Laboratory for Powder Metallurgy, Central South University, Changsha 410083, China

M AN U

SC

b

Abstract

The structures and energetics of η′ nano-phase and its interfaces in a peak-aged Al-Zn-Mg-Cu alloy were thoroughly investigated, using the combination of aberration corrected HAADF-STEM imaging and first-principles calculations. The most feasible atomic

TE D

structure of η′, along with the solute substitution in η′, were calculated and compared with the atom resolution Z-contrast images. The interface phase diagram of η′/Al was constructed as a function of the excess chemical potential of Zn, to determine the equilibrium interface

EP

structures. Solute segregation to these interfaces was further calculated, and the results were adopted to interpret the experimental Z-contrast images. Finally, all the bulk and interface

AC C

results were integrated to predict the solute partition in the matrix and further its potential impacts on interface properties, and based on which, a new strategy was proposed for future optimal design of Al-Zn-Mg-Cu alloys. Keywords: Al-Zn-Mg-Cu alloy; η′; interface; HAADF-STEM; first-principles

∗

Corresponding author. School of Materials Science and Engineering, Central South University, Changsha, 410083, China ∗∗ Corresponding author. School of Materials Science and Engineering, Shanghai Jiao Tong University, Shanghai 200240, China E-mail addresses: [email protected] (Y. Jiang), [email protected] (Bin Chen). 0

ACCEPTED MANUSCRIPT

1. Introduction Al-Zn-Mg-Cu alloys are widely employed in many light-weight structural applications due to their high specific strength, good corrosion resistance and workability, and relatively

RI PT

low production cost. The strengthening of the alloys is achieved mainly from the dispersive precipitation of high density nano-size second-phases from the supersaturated solid solution through aging treatments. Among various nano-phases, metastable phase η' is generally

SC

believed to be responsible for the maximum strengthening at peak aging [1-3]. Given its significance, extensive research efforts have been devoted to characterize the atomic structure

M AN U

and composition of η'. The generic precipitation sequence of η' under artificial aging has been suggested as: super-saturated solid solution (SSSS)→GP zone→η'→η(MgZn2) [4,5], but the proposed composition of η' is very diverse in literatures, ranging from MgZn2 [6,7],

TE D

Mg2Zn3Al4 [8, 9], Mg2Zn5-xAl2+x [10], Mg2Zn5Al2 and Mg2Zn7Al [11] to Mg4Zn11Al [12]. Though the hexagonal crystal with lattice parameters of a=b=~4.96 Å and c=14.02~14.05 Å is more often adopted for describing η' phase, there are many controversies about its structural

EP

details, including the lattice parameters, internal atomic coordinates, and elemental site

AC C

occupations [5,8,13-15]. Historically, such discrepancies can be majorly ascribed to two reasons: (1) the technical difficulties in direct characterizing nano-sized precipitate phases with a complex composition in a host lattice, and (2) the fact that various transition structures during the transformation from GP zones to the stable phase η-MgZn2 are all termed as η'. Coherent GP zones, including both the spherical GPI (featured with a local ordering of Zn and Al/Mg atoms on the {001}Al planes) and the disk-like GPII (featured with one Zn-rich {111}Al plane), have been suggested to be closely related to the formation of η', but which one

1

ACCEPTED MANUSCRIPT contributes more remains uncertain [5,16,17]. In addition, the stable phase η presents nine different orientation relations of η1~η9 with the Al matrix [18]. η' has an orientation relation of (0001)η′//{111}Al and (10-10)η′//(110)Al in Al which is almost identical to η2, and thus has

RI PT

been specifically proposed as the precursor of η2 [19]. Meanwhile, a new transition structure, ηP, has been also suggested as the precursor of η [20]. Although ηP-MgZn2Al has the same orientation relation and a similar morphology as η', Liu et al [20] proposed to categorize it

SC

separately from the η' class, for its c parameter (c=9.35Å) seriously deviates from η' (14.02~14.05 Å) but is more comparable to η-MgZn2 (8.60 Å). The a,b parameters are yet

M AN U

close among η', ηP, and η, i.e. 4.96~5.21 Å.

The ever-increasing demanding for light-weight high-strength Al–Zn-Mg-Cu alloys from aerospace and civil transportation industries consistently drives the further studies of η'

TE D

nano-phase precipitates towards the optimal design of the alloys. Besides many uncertainties about the bulk structure of η', we have even less knowledge about its interfaces with the Al matrix. Both are equally critical for developing a fundamental understanding of η' formation

EP

and behaviors in Al alloys [21-23]. For example, interface energy, defined as the free energy

AC C

cost in forming new interface areas in Al matrix, are combined with bulk energies to determine the critical free energy of nucleation and the critical radius of nucleus, and further the volumetric fraction of precipitates in Al. Also, segregated elements may strongly affect interface energy by changing the local chemical bonding at interface, and hence the thermal growth and size stabilities of precipitates [24]. Furthermore, the anisotropy in interface energy directly governs the morphologies of precipitates [25,26]. Undoubtedly, an intricate understanding of the structures and energetics of η' phase and its interfaces will provide

2

ACCEPTED MANUSCRIPT important insights into all the above fundamental aspects and their complex interplays. The multi-element composition of η', along with its nanoscale size and the associated local strains, has historically posed arduous challenges for atom-level characterization of η'

RI PT

/Al interfaces. Conventional transmission electron microscopy (TEM) images lack sufficient spatial and elemental resolution for this purpose. By using high resolution TEM (HRTEM), it has been revealed that η' precipitates possess a thin disc-like morphology with an orientation

SC

relation of (0001)η′//(111)Al and (10-10)η′//(110)Al in Al matrix [27]. The phase contrast images obtained via HRTEM, however, inevitably contain delocalization information of

M AN U

atoms, and cannot be applied to precisely locate the atom positions in η' bulk phase. Scanning TEM (STEM) in high-angle annular dark-field (HAADF) imaging mode can provide more interpretable atomic resolution images (i.e. so-called Z-contrast image) in which the contrast

TE D

of atomic columns is proportional to the square of the atomic number Z. However, except for the orientation relationship and the high Z-atom enriched double layer of η' in Al [19], there have been no HAADF-STEM efforts devoted to interpret the atomic structure and

EP

composition of η'/Al interfaces yet. In literatures, only a few HAADF-STEM studies have

AC C

been focused on the η/Al interfaces, but nevertheless, only specifically on the broad interface of {0001}η/{111}Al. For instance, one early HAADF-STEM study of an Al-8.5Zn-1.75Mg2.3Cu alloy suggested that disc-shaped η elongates on {111}Al planes and exhibits some distinct compositional ordering: Mg- and Zn-rich atomic layers alternate in the η bulk interior, and the broad η/Al interface is terminated with Zn-rich {0001} layers [28]. However, the Z-contrast difference between neighboring elements in the periodic table, i.e. between Mg and Al or between Zn and Cu, is literally imperceptible on HAADF-STEM images. Thus it is still

3

ACCEPTED MANUSCRIPT inadvisable to rely solely on electron microscopy techniques to explore all fundamental aspects of such complex nano-phases in Al alloys. By combining the aberration-corrected STEM with first principles calculations on an Al-8.5Zn-2.2Mg-1.9Cu alloy, Marioara et al.

RI PT

[29] were able to identify the outmost terminating layer of η phase to be a bi-layer structure. The two sub-layers are slightly separated, with the outer sub-layer being enriched in Zn and the inner one enriched in Mg. Clearly, both HAADF-STEM studies agreed on the Zn-rich

SC

outmost layers of η in Al, but first-principles calculations were able to provide more insights into the structural details, and beyond which, the relevant energetics and properties of the

M AN U

interfaces. More recently, Tsurua et al. [30] suggested the possible hydrogen trapping sites both inside η and at the η/Al interface solely using first principles calculations. In fact, characterizing such light atoms in such complex structures has been beyond the general

TE D

capability of current electron microscopy techniques.

Enlightened by the previous studies on the η/Al interfaces, we propose to couple the aberration-corrected HAADF-STEM analysis with the first-principles density functional

EP

theory (DFT) calculations in this work, to investigate the fundamental aspects of η' bulk and

AC C

its interfaces in Al. The compositions and energetics of various bulk structures of metastable phase η' proposed in the literatures were first examined by DFT calculations, to determine the most energy-favored one. Further based on the atomic resolution STEM, the structures and energetics of η'/Al interfaces were modeled and calculated. An interface phase diagram was constructed to determine the thermal equilibrium interface structure and termination chemistry, for a wide range of Zn chemical potentials. Also, solute substitution inside η' bulk and further solute segregation to the interfaces were thoroughly evaluated, to help interpret

4

ACCEPTED MANUSCRIPT the experimental Z-contrast images. Based on all the results, their possible implications for stabilizing η' nano-precipitates in Al-Zn-Mg-Cu alloys can be suggested.

RI PT

2. Methods 2.1 Experiments and simulations

An commercial 7075 Al alloy with a composition of 5.47% Zn, 2.54% Mg, 1.44% Cu,

SC

0.45% Fe, 0.33% Si, and 0.15% Ti (all in wt.%) was adopted. The alloy ingot was solution treated at 500℃ for 3h followed by water quenching. Fig. 1 shows the age-hardness curve of

M AN U

the alloy aged at 150℃. As shown, the hardness reaches its peak value of ~190 HV after 6~8 h of ageing. The quenched alloy was then preserved in silicon oil and aged at 150℃ for 6h to peak aging [9]. Disk samples with a diameter of 3 mm and a thickness of ~50 µm were

TE D

punched from manually ground slices cut from the aged ingot, and twin jet electro-polished at -35 ℃ with a 1/3 HNO3+2/3 methanol solution. Atomic resolution HAADF-STEM characterization was carried out on a JEM-ARM200F, equipped with a probe Cs-corrector

EP

and a cold field emission gun at 200 kV. The camera length was set to 8 cm, yielding a

AC C

collection semi-angle of 68∼280 mrad. The precipitate images were taken along the <112>Al to achieve a better view of inter-columnar separations with the limited overlap with Al matrix in this projection.

5

SC

RI PT

ACCEPTED MANUSCRIPT

M AN U

Fig. 1 The age-hardness curve of the alloy during isothermal aging at 150℃.

HAADF-STEM image simulations were performed to compare the computed structure model with experimental result. The HAADF-STEM image simulations were conducted using

TE D

multi-slice calculations incorporating thermal diffuse scattering via a frozen phonon approach by QSTEM. The simulations were performed for atomic models within supercells of lateral size 50Å×50Å, sampled using 400×400 pixels. The calculations assume an effective source

EP

size of 0.7Å .

AC C

2.2 DFT Calculations

All structural relaxation and energetic calculations were performed using the DFT code VASP (Vienna Ab Initio Simulation Package) with periodic conditions and the plane-wave basis sets [31]. The electron-core interaction was described by the Blöchl projector augmented wave method (PAW) within the frozen-core approximation [32, 33]. Validation of the exchange-correlation (XC) functional was performed by fitting the energy-volume relation for fcc Al and hexagonal η' to the universal equation of state [34], to reproduce the

6

ACCEPTED MANUSCRIPT experimental lattice constants. The tested XC-functionals included local density approximation (LDA) of Ceperley and Alder [35, 36], the generalized gradient approximation (GGA) of Perdew-Wang (PW91) [37] and the Perdew-Burke-Ernzerhof (PBE) functionals

RI PT

[38]. The LDA was finally adopted for it yielded the best prediction for η' (a=4.98Å, c= 14.25 Å) as compared to the experiment values of a=4.96Å, c=14.02 Å [27, 39]. Other validations based on elastic properties and even electronic structures are not feasible yet, owing to the

SC

shortage of corresponding experimental results. A sufficiently high energy cutoff of 350 eV was adopted for the plane-wave basis set expansion. For all the considered interface structures,

M AN U

we used a sandwich supercell of Al/η′/Al with a 2×3×1 Monkhorst-Pack k-mesh. Full relaxation was performed on both the supercell volume and shape, to optimize the ground-state atomic geometries by minimizing the Hellman-Feyman forces until the total

TE D

forces on each ion were converged to within 0.02 eV/Å.

3. Results and discussion

EP

3.1 HAADF-STEM images of η' in Al

AC C

The peak aging treatment (T6 state) leads to mostly η' precipitates in Al-Zn-Mg(-Cu) alloys [3, 9, 40]. Fig.2a shows the dispersive distribution of nano-size η'-phases precipitated from the peak-aged specimen as viewed along the <112>Al. Fig. 2b shows the typical morphology of η' precipitates in the matrix. Further viewing along the <111>Al confirmed these η' nano-particles have a thin disk-like morphology. They tend to align parallel to {111}Al planes, with a rather uniform thickness of 5∼6 nm and a more diverse diameter ranging from ~10 to ~20 nm (see Fig. 2). The high density dispersion of η' nano-precipitates

7

ACCEPTED MANUSCRIPT contributes mainly to the peak strengthening of the alloy [9]. The determination of these nano-precipitates as η' rather than η is further supported by the low nanometer size, the thin disk-like morphology, the smallest repeatable unit cell size, and more importantly, the best

RI PT

match achieved between the HAADF-STEM atomic images and the most energy-favored η'-phase structure predicted by our first principles calculations (as we will introduce in the next section). The size and morphological features also easily distinguish η' phases from

SC

GPII zones [20].

η' interfaces with the Al matrix through two basic types of contacting facets. The broad

M AN U

interface is coherent and has been identified to be the {0001}η’/{111}Al, while the periphery interfaces are semi-coherent and has been suggested to be mainly the {10-10}η’/{110}Al [9]. Perhaps due to the generally circular-shaped periphery boundary, we failed to obtain a

TE D

sufficiently high resolution image of the periphery interface {10-10}η’/{110}Al. Fig. 2c shows the atomic resolution Z-contrast image of the board interface {0001}η′/{111}Al. Clearly, the broad interface is fairly flat and smooth, and possesses a high coherency with the

EP

Al matrix lattice. A closer examination on the coherent broad interface reveals that the η'

AC C

phase half is terminated with a series of bright atom pairs (the B-layer) and the Al lattice half is terminated by a layer of darker atomic columns (the A-layer). On the Z-contrast images of an Al-Zn-Mg-Cu alloy, the brighter contrasts shall denote Zn or Cu-rich atomic columns and the lower contrasts denote Al or Mg atomic columns. Thus one can deduce that the terminating B-layers of η' is somewhat enriched by Zn or Cu. Also note that the B-layer atomic contrasts are not so bright and uniform, and the middle region become darker in Fig. 2c, indicating the possible partial substitution of Zn or Cu on the interfacial B-layer (as we

8

ACCEPTED MANUSCRIPT will further discuss below). Although the Z-contrast HAADF-STEM images clearly show the board interface boundary, it still lacks the elemental information that is sufficient to distinguish between Zn and Cu, or between Al and Mg. Moreover, whether the bi-layer

RI PT

feature previously proposed for the board η/Al interface also presents for the terminating B-layer of the board η'/Al interface remains unclear. To address these problems, we resorted

EP

TE D

M AN U

SC

to the first-principles calculations.

AC C

Fig. 2 (a) A low magnification image of nano-sized η' precipitates in the peak-aged specimen as viewed along the <112>Al, (b) the morphology of a typical η' precipitate in the matrix, and (c) the atomic resolution Z-contrast image of η' phase and its board interface with Al. The interface has the typical orientation relation of (0001)η′//(111)Al and (10-10)η′//(110)Al

3.2 Bulk structure of η' and solute substitution Several crystal structures have been previously proposed for η' in literatures and are now

9

ACCEPTED MANUSCRIPT summarized in Table1. These structures are distinct from one another in terms of composition, the Zn/Mg atom ratio, and the lattice parameters as well. Multiple atom probe tomography (APT) studies have suggested that the Zn/Mg ratio of these nano-precipitates in Al-Zn-Mg-Cu

RI PT

alloys generally falls within a range of 0.9∼1.8 [5, 13, 41-44]. Mg4Zn13Al (or its analog, Mg4Zn11Al [12]) can be thus excluded for η', due to its apparently too high Zn/Mg ratio (of ~3.25), although it exhibits relatively low formation energy [15]. For the same reason,

SC

Mg2Zn7Al, Mg2Zn5Al2 [11] and MgZn4Al [45, 46] can be also excluded. Furthermore, MgZn2 was also highly questionable due to the generally accepted finding that solute Al enters into η'.

M AN U

Bearing all these in mind, we felt comfortable to focus on three model structures of η' only, i.e. Model η'-I and II proposed by Kverneland et al [8] and Model η'-III proposed by Li et al [27] after structural refinement [47]. The three models have the same nominal composition of

TE D

Mg2Zn3Al4 but with different internal atomic coordinates and lattice parameters. The relative stabilities of the three model structures can be assessed by calculating the formation energy per average atom as

’

’

− x

+y

+z

/(x + y + z)

(1)

is the total energy of the η'-Mg Zn Al supercell. µi is the chemical potential of

AC C

where

EP

∆E =

Mg, Zn, or Al in the alloy matrix. The chemical potentials can be approximated by evaluating the total energy difference between the dilute solution of Al107Z and pure Al using the same 3×3×3 fcc supercell, i.e.

(Z) = E(Al#$% Z) − 107 (Al). The chemical potentials of Mg and

Zn in Al are thus calculated as -1.64 and -1.76 eV, respectively. For a more accurate prediction for elemental chemical potentials, one may refer to one our earlier work [48].

10

ACCEPTED MANUSCRIPT Table.1 Survey of the crystal structures proposed for η’ phase Structure

Lattice constants

Nominal composition

Hexagonal

a=4.96Å,c=8.68Å

MgZn2

Notes Proposed by Mondolfo al[6]. The first hexagonal structure proposed for η’

A=4.97Å,c=5.54Å

MgZn2

Proposed by J. Gionnes[7], but with an

RI PT

Orthorhombic

unfavorably high formation enthalpy[15] Hexagonal

a=4.96Å, c=14.02Å

Mg4Zn11Al

Proposed by Auld [12]. Supported by DFT calculations and can be optimized to Mg4Zn13Al [15], but have been challenged by others [5, 27]

a=4.96Å, c=13.92Å

Mg2Zn7Al/

Two models with different composition are

Mg2Zn5Al2

proposed by H. B. Larsen [11] through X-ray

SC

Hexagonal

Hexagonal

a=4.96Å, c=14.05Å

M AN U

diffraction data analyses.

Mg2Zn3Al4

Two models (denoted as Model I and II) with different atomic coordinates as proposed by Kvernel and [8]. Model I better matches the HAADF results by Xu et al.[9]

Hexagonal

a=4.96Å, c=14.02Å

Mg2Zn5-xAl2+x

Proposed by Li et al [27]. Can fit into the composition of Mg2Zn3Al4 after structural

a=5Å, c=9.27Å

MgZn4Al

Proposed by Liu et al.[45, 46]. Considered as ηp, a transition structure from η’ to η

EP

Hexagonal

TE D

refinement [47] (denoted as Model III)

AC C

The calculated formation energies of the three model structures of η' are compared with both the stable phase η and the transition phase ηp in Fig. 3a. Positive formation energy predicts an unfavorable precipitation from the solid solution. Clearly, the η'-II structure has a positive formation energy and thus is not favorable. The η'-I structure is slightly more energy favored than the η'-III. The equilibrium phase η has the lowest formation energy, followed by the transition phase ηp. All the considered η' model structures are predicted to be less stable than ηp and η, and the corresponding energy differences contribute to the thermodynamic

11

ACCEPTED MANUSCRIPT driving forces for metastable η' to transform towards the more stable phases ηp and η. Moreover, by comparing the lattice parameters of the three η' models and the fcc-Al, one can deduce that the η'-I structure is under relatively smaller mismatch strains in the Al matrix,

RI PT

with 0.95% along [10-10]η' and 1.5% along [0001]η', as compared to the 4.46% and 1.1% for the η'-II and the 3.26% and 1.09% for η'-III, respectively. This suggests a better elastic strain compatibility between η'-I and the Al matrix, in favor of the η'-I formation.

SC

Fig. 3b superimposes the atomic-resolution HAADF-STEM image of η′ bulk interior with the fully-relaxed atomic structure and the corresponding HAADF-STEM simulation

M AN U

image of η'-I. Note that the unit cell of η' (indicated by a red frame) is assembled by two abreast hexagonal lattices under inversion symmetry. The locations of Zn atoms in the η'-I structure matches well with the bright dots on the atomic Z-contrast image. Also note that the

TE D

pair of Zn columns in the unit cell of η' presents different contrasts along the <112>Al. In fact, one column of the pair has twice numbers of Zn atoms as the other column (see the inset in Table 2) and thus appears to be always brighter. Al and Mg columns are nearly

EP

indistinguishable on the Z-contrast image, as one can expect. There are also some unexpected

AC C

bright dots appearing at the locations which are supposed for Al atoms. These bright dots are aperiodic and occur very occasionally inside the η' lattice, possibly due to Zn or Cu substitution as we will discuss below.

12

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

Fig. 3 The bulk structure and energy of η' phase. (a) Calculated formation energies of the three model structures of η'-Mg2Zn3Al4 in comparison with the transition phase ηp-MgZn2Al and the stable phase η-MgZn2 (the open square and circle are taken from other calculation results [15, 46], respectively). (b) An enlarged view of the η' bulk interior (inside the dashed box in Fig.2b) being super-positioned with the fully relaxed structure and the corresponding simulation image of η'-I.

13

ACCEPTED MANUSCRIPT Previous studies have suggested that the compositions of η' precipitates in Al-Zn-Mg-Cu alloys can hardly be perfect stoichiometric, but vary with the alloy composition and the heat treatment history [5, 42, 49]. To assess the solute dissolution in η' bulk, we calculated the

∆

)*+

=,

η0 -→/

−

η0

1−(

-

−

/

RI PT

substitution energies for each solute species as ),

(2)

where E-→/ is the total energy of η' after j atom being substituted by i atom, and E η0 is the η0

SC

total energy of perfect η'. Both the two terms were calculated using a 2×2×1 supercell of η' with a 3×3×2 Monkhorst-Pack k-mesh. μ-// is the chemical potential of element i

M AN U

(=Mg/Zn/Cu) or j (=Mg/Zn/Al) in the alloy matrix, as we have calculated above. The calculated substitution energies are compared in Table 2. A negative ∆Esub value predicts a feasible substitution path from the matrix to η' bulk interior. Noticeably, Cu substitution

TE D

energies at all the selected sites in η’ were calculated as negative, and especially, the Zn1 site is the most preferential site for Cu. Cu also shows a strong preference to the Al2 and Mg sites. Meantime, solute Zn exhibits only a weak preference to the Al1 sites. These results seemly is

EP

in accordance with the previous time-resolved anomalous small-angle X-ray scattering

AC C

(ASAXS) study on a series of aged Al-Zn-Mg-Cu alloys [50]: the Zn concentration of nano-precipitates decreases steadily while the Cu concentration increases correspondingly with aging time, especially for higher Cu content alloys aged at higher temperatures. Solute Mg also strongly prefers the Al1 and Al2 sites. Considering the very similar Z-contrasts between Mg and Al, the unusual bright contrasts at some Al sites in Fig. 3b must be majorly attributed to Cu substitutions. Furthermore, the strong preference of Mg substitution at the Al sites in η' also suggests that the high Al-content η'-Mg2Zn3Al4 may not be stable, but have the

14

ACCEPTED MANUSCRIPT tendency to transform into the more stable Al-lean phase ηp (MgZn2Al), and further the Al-free stable phase η (MgZn2). This may also help explain how η' evolves into η through the transition phase ηp. This also consists well with the previous atom probe tomography (APT)

RI PT

and SAXS studies on the Al-Zn-Mg alloys [14, 51]: during over aging of the alloy, nano-precipitates change from mostly η' to mostly η, and accordingly, the Al concentration in nano-precipitates decreases while the Mg concentration increase consistently with aging time.

SC

Noticeably, the addition of Cu with limited amount does not affect this evolution trend. As experimentally revealed for many Al-Zn-Mg-Cu alloys, up to 2.5 wt% Cu can dissolve in η'

M AN U

and η phases without changing the precipitation sequence during aging [52, 53].

Table 2. Predicted substitution energies of solute atoms in η’ phase Substituted site

Mg

Cu

Al1

-0.11

-2.45

-0.01

Al2

0.09

-2.73

-2.41

Mg

0.47

-

-1.63

Zn1

-

0.60

-3.32

Zn2

-

0.60

-0.10

AC C

EP

Substitution energy (eV/atom) Zn

TE D

Atomic structure of η’

3.3 Interface structures and energetics Recall that η' nano-precipitates present a disk-like morphology in Al, particularly with the broad face (0001) interfacing with the Al{111} planes and the periphery face {10-10} interfacing with the Al{110} planes. Both the two basic interfaces are studied here. To model the coherent broad interface of (0001)η’/(111)Al, we compressed the η' lattice by 2% along 15

ACCEPTED MANUSCRIPT [10-10]η' (//[110]Al) to achieve the perfect lattice matching of d(10-10)η' with 3d(220)Al. This leads to a very small commensuration strain of ε(η')=-0.95% along [10-10]η'. To model the semi-coherent peripheral interface of (10-10)η′/(110)Al, it is still not feasible to naturally

RI PT

include the misfit dislocations into our supercell model. We chose to compress the η' lattice by 3.4% along [0001]η' (//[111]Al) to achieve an overall matching between d(0001)η' and 6d(111)Al. This leads to a minimal commensuration strain of ε(η') =-1.50% along [0001]η'.

SC

Both the resulting structural ensembles are in accordance with the previous HRTEM observations[27]. The interface ensembles were then optimized upon full structural

M AN U

relaxations, to determine the ground-state atomic coordinates in the near-interface region. Fig. 4a schematically shows the orientation relation of η’ in Al. By carefully reviewing the stacking characters of η'(10-10) and η'(0001) surfaces, we proposed four possible termination

TE D

types for the broad interface {0001}η′/{111}Al, namely the Al2-, Zn-, Mg-, and Al-terminated, and three possible termination types for the periphery interface {10-10}η'/{110}Al, namely the stoichiometric AlZn-terminated, the non-stoichiometric Al-terminated and MgAlZn-

EP

terminated. Fig. 4b compares the constructed supercell models of η'/Al interfaces with all

AC C

various termination types. The termination type directly determines the chemical stoichiometry of the interface. Besides the orientation relation and interfacial termination (or stoichiometry), interfacial atomic coordination type also dictates the local chemical environment. These feature parameters play a combined role in determining the interface structure and properties, including formation energy, adhesion strength, elemental segregation, and more [54,55]. Thus for each termination type, we further considered three different coordination types by translating the Al matrix relative to η' phase, i.e. placing the Al atoms

16

ACCEPTED MANUSCRIPT right above, at the bridge sites, or hollow sites of the surface Mg or Zn atoms, namely the top-coordinated, the bridge-coordinated, and the hollow-coordinated (not individually shown in Fig. 4b). First-principles calculations were then performed on a total of 21 model structures,

AC C

EP

TE D

M AN U

SC

RI PT

to determine the most energy-favored interface structures.

17

ACCEPTED MANUSCRIPT Fig. 4 (a) A schematic 3D drawing to show the orientation of a disk-like η′ nano-precipitate in the Al matrix. (b) The constructed sandwich supercell models for the broad interface {0001}η’/{111}Al and the periphery interface {10-10}η’/{110}Al with various possible

3.3.1 Interface formation energy and interface phase diagram

RI PT

interfacial termination types.

#

2=3

E454 − N

SC

A generalized definition of interface formation energy can be expressed as [54, 56] −N

+ 7ΔV − TΔS ,

−N

(3)

M AN U

where Etot is the total energy of a fully relaxed interface, A is the cross-sectional area of the interface, µi is the elemental chemical potential in the alloy matrix, and Ni is the number of the corresponding atoms in the interface supercell. ∆V is the volume change due to interface

TE D

relaxation. ∆S is mainly the vibrational entropy difference between the interface and the alloy matrix. T is the aging temperature within a general range of 100~200°C, far below the melting point (∼660°C). The last two terms shall be relatively small and thus can be disregarded.

EP

Only the contributions from the former energy terms need to be calculated.

AC C

When the η'/Al interface reaches its thermal equilibrium, the sum of the elemental chemical potentials, ∑Niµi, is equal to the chemical potential of η' (Al4Mg2Zn3), µη', i.e. ∆E=0 for η' formation or decomposition. As a pure substance,

0

=

5 0,

where the

superscript o stands for the standard state. We thus have 4

+2

+3

=

0

=

5 0,

Alternatively, the elemental chemical potentials at the standard state, with,

5 0,

(4) 5 - ,,

can be also related

by the formation energy of η’ phase in Al, ∆HC5 , 18

ACCEPTED MANUSCRIPT 4

5

5

+2

+3

5 D

+ ∆HC5 =

5 0,

(5)

By defining the excess chemical potential for element i at the interface, ∆

-

=

-

−

5 -,

and

subtracting Eq. (5) from Eq. (4), one has = ∆HC5 .

+ 3∆ ≅

Here, for a dilute Al-based solid solution,

5

[48]. Combining Eqs. (3) and (4), one

can rewrite the interface formation energy to be #

3

E454 − (N − 2N

5

− ,N

F

− N 3

1

5

#

− N 3

5 0

− ,N

SC

2=

(6)

RI PT

2∆

M AN U

For a given temperature, each chemical potential (unit energy) term

5 -

F

− N 3

1∆

]. (7)

holds as a constant,

but the relative atom numbers and hence Etot in Eq. (7) varies among different interface structures. Due to large cancellation occurring in Eq. (7), as the first order approximation, all 5 -

can be estimated by 0-K enthalpy calculations. Thus the

TE D

the total energy terms Etot and

temperature dependence of interface energy relies dominantly on ∆µZn (or µZn). Please note, ∆μ

= kTlnR , reflecting a combined effect of temperature (T) and the chemical activity of

EP

Zn (aZn). The aZn itself, by definition, is proportional to the atom concentration of Zn through

AC C

the activity coefficient γZn which is also a function of temperature. To ensure a stable interface, the chemical potential of Mg or Zn at interface must be limited to less than that in its pure bulk standard state, thus ∆µZn <0 and ∆µMg <0. Combining with Eq. (6), the reasonable varying range of ∆µZn can be determined as # F

∆HC$ ≤ ∆μ

≤ 0.

(8)

The formation energy of bulk phase η', ∆HC$ , is calculated to be -6.92 eV per formula. Using Eqs. (7) and (8), we can calculate interface energy as a function of ∆μ

and plot all the

19

ACCEPTED MANUSCRIPT results in Fig. 5. Again, ∆μ

measures the chemical activity and

As suggested by Eq. (7), it is the Zn/Mg ratio of η' that determines how sensitively a non-stoichiometric interface energy would depend on ∆μ

(or µZn). Clearly in Fig. 5, the

RI PT

Zn-terminated (Zn-rich) broad interface energy decreases while the MgZn-terminated (Mn-rich) periphery interface energy increases as increasing ∆μ

or µZn. A stoichiometric

interface with a Zn/Mg ratio of 2/3 has a constant interface energy independent of ∆μ , such

in most of the ∆μ

SC

as the Al-terminated BI. The stoichiometric Al-terminated BI has the lowest formation energy range. Only at very high Zn potentials, i.e. ∆μ

≥ ~ − 0.15 eV, could

M AN U

the Zn-terminated (Zn-rich) BI become the most stable structure. As for the {10-10}η′/{110}Al PI, the stoichiometric AlZn-terminated structure is always the most energy favored till the μ

is reduced to a very low value, i.e. ∆μ

= ~ − 1.8 eV. Below which, the most

energy-favored PI structure may change to the MgZn-terminated. We notice that the

TE D

formation energy of {0001}η′/{111}Al BI can be extremely low, ~44m J/m2 for the Al-terminated BI, while the periphery interface of {10-10}η′/{110}Al has a much higher

EP

formation energy, 190 mJ/m2. This suggests the best strategy for η' precipitation in Al: η'

AC C

nucleates preferentially on the {111}Al and then grow continuously along the <110> Al, which eventually results in a disk-like morphology along the {111}Al in Al. It is also nature to deduce that the disk diameter shall increase ahead of the thickness during η' growth, so as to minimize the enengy cost before η' reaches a given volume size in Al. This agrees well with many high-resolution microscopic observations on the growth of η' nano-disks during the continuous aging [9, 28]. We are not aware of the realistic Zn potentials at the local interface region, but there is highly possibly a high degree enrichment of Zn on the {111}Al of the

20

ACCEPTED MANUSCRIPT as-nucleated precipitates as we previously suggested using atomic-scale HAADF-STEM [9], thus the Zn potential in the alloy matrix can be rather high. Once the Zn potential falls into the range of ∆μ

≥ ~ − 0.15 eV, according to the calculated interface phase diagram in Fig. 5,

RI PT

η' nano-disks tend to possesses a Zn-terminated (Zn-rich) BI together with a stoichiometric AlZn-terminated PI. Futhermore, based on the calculated interface energy values, we can estimate the interface anisotropy (between the broad face and the periphery face) and hence

EP

TE D

M AN U

SC

the equilibrium aspect ratio of η' nano-disks to be ~4.3, by following the method in Ref. [57].

AC C

Fig. 5. The calculated interface phase diagram to determine the most energy favored atomic structures for the {0001}η’/{111}Al broad interfaces (BI) and the {10-10}η’/{110}Al periphery interfaces (PI).

3.3.2 Atomic interface structures To further ascertain the atomic occupation at the broad interface of {0001}η’/{111}Al, we placed the DFT-optimized interface structures as well as the corresponding HAADF-STEM 21

ACCEPTED MANUSCRIPT simulation image on top of the atom resolution Z-contrast image (Fig. 2c) , to search for the best overlapping in Fig. 6. Clearly, only the Al-terminated interface (see Fig. 6a) achieves an almost perfect match between the calculated atomic structure and the atom resolution contrast

RI PT

image, on both the Al and the η′ sides of the interface. On the η′ side, the Zn atoms well overlap with the bright columns while Mg and Al atoms overlap with the lower contrast columns. The Al2-terminated interface in Fig. 6b, however, misses the overlapping at the

SC

outmost layer of the Al half (i.e. the A-layer defined in Fig. 12c). Meanwhile, pronounced atom-contrast mismatch is observed on the η′ side for the Zn-terminated interface case (Fig.

M AN U

6c), and on both sides for the Mg-terminated interface case (Fig. 6d). Recall that in Table 2, we have predicted the high possibility of Al substitution with Mg (or Zn substitution with Cu) inside η′, but unfortunately, the very similar atomic numbers (Z) of Al and Mg (or of Zn and

TE D

Cu) make such substitution nearly indistinct on the HAADF-STEM Z-contrast images. To fully manifest on this, atom resolution EDS mapping and analyses shall be desirable in future studies. Table 2 also reveals the strong tendency of Al2 or Mg substitution with Cu in η′. The

EP

possible sites of Cu substitution in η′ are indicated using yellow arrows in Fig. 6a. Complete

AC C

substitution of a whole column can hardly occur, and it shall be more realistic that partial substitution takes place occasionally inside these atom columns, and as a direct consequence, the substitution-induced contrasts vary much among these sites along the <112>Al. According to Table 2, we cannot exclude the weak possibility of Al1 substitution with Zn in η′, but both the substitution energy in Table 2 and the contrast image in Fig. 5a suggest such substitution to be very limited. Fig. 4 predicts the high stability of the stoichiometric Al-terminated BI in a wide Zn potential range. Only at very high level of µZn (∆μ

≥ ~ − 0.15 eV), the most 22

ACCEPTED MANUSCRIPT energy favored structure could change to the Zn-terminated one. The interface termination in Fig. 6a is obviously not the Zn-terminated but possibly has been segregated with a few high-Z atoms (Zn or Cu, as indicated by white arrows in Fig. 6a). Minor interfacial segregation of Zn

RI PT

is in accordance with the relatively low Zn content (~5.47 wt%) in our sample alloy, and perhaps also with the low diffusivity of Zn at the relative low aging temperature (150℃). Further increasing the Zn content in the alloy matrix and/or the aging temperature may result

SC

in a higher potential of Zn which shall favor the formation of the Zn-terminated BI. The former approach will be thoroughly examined in the next section. Further, based on all above

M AN U

results and discussions on the broad interface, the Zn potential must be intermediately high, and thus we feel comfortable to predict the periphery interface {10-10}η’/{110}Al to be the

AC C

EP

TE D

stoichiometric AlZn-terminated, with a higher interface energy of 190 mJ/m2.

Fig. 6. The HAADF-STEM Z-contrast image of the broad interface {0001}η’/{111}Al (the same view of Fig. 2c) superimposed with DFT-calculated interface structures with four 23

ACCEPTED MANUSCRIPT different terminations: (a) the Al-terminated, (b) the Al2-terminated, (c) the Zn-terminated, and (d) the Mg-terminated.

RI PT

3.3.3 Interface segregation and solute partition So far, our interface calculations have been targeted on clean η′/Al interface. The occasional bright dots observed on the B-layer of the Al-terminated BI suggest the possible

SC

interface segregation of high-Z solute atom Zn and/or Cu (indicated by white arrows in Fig. 6a). We thus proposed to examine the potential solute segregation of Zn, Cu, and Mg from the

M AN U

alloy matrix to the two equilibrium interfaces, i.e. the Al-terminated BI and the AlZn-terminated PI, by calculating the segregation energy as )c

ΔE)c = E454 − E454 − (μ- − μ/ )

(9-a)

)c

where E454 is the total energy of the clean interface. E454 is the total energy of the

TE D

segregated interface. μi/j is the chemical potential of replaced atom j (=Mg, Zn, Al) and the replacing solute atom i (=Mg, Zn, Cu) in the alloy matrix. Negative segregation energy

EP

corresponds to an energy favored segregation path. Especially, for solute segregation to an

AC C

interfacial interstitial site, the segregation energy can be rewritten as )c

ΔE)c = E454 − E454 − (μ −μ- )

(9-b)

Solute segregation to various sites at the {0001}η’/{111} broad interface and the {10-10}η’/{110}Al periphery interface are calculated and compared in Fig. 7. Clearly, solute segregation depends sensitively on the interface structure as well as the solute species. For the {0001}η′/{111}Al broad interface, all the solute species prefer, and only prefer, the interfacial Al sites, and more specifically, prefer Al1 sites to Al2 sites (as denoted in Fig. 7a). This

24

ACCEPTED MANUSCRIPT conforms to the few bright spots observed on the B-layer rather than the A-layer of the interface. On the {10-10}η′/{110}Al periphery interface, all the interfacial Al sites attract solute Mg, and many of them also slightly attract Zn. Cu mainly segregates to the interstitial

RI PT

sites. It is obvious that, due to the structural discrepancy between the interface and the two bulks, segregation-induced interfacial substitution behaviors totally different from the bulk counterparts in Table 2. The Mg and Zn1 sites in η′ bulk are highly favorable for Cu

SC

substitution (see Table 2), but the interfacial Mg and Zn sites at both interfaces are not. Also, atomic radius is not the decisive factor in determining the substitution behaviors, and instead,

M AN U

atomic electronic structure often plays the critical roles. For instance, the interstitial sites on the broad interface do not attract any solutes, including Mg and even the smaller sized Cu and Zn. While on the periphery interface, the interstitial sites are only attractive to Cu but not to

AC C

EP

TE D

the similar-sized Zn.

25

SC

RI PT

ACCEPTED MANUSCRIPT

M AN U

Fig. 7. Segregation energy of solute Zn, Cu, and Mg from the alloy matrix to (a) the broad interface of {0001}η′/{111}Al and (b) the periphery interface {10-10}η′/{110}Al.

Solute atoms Zn, Cu, and Mg may dissolve into the η′ bulk interior by substitution (as

TE D

revealed in Table 2 and suggested by some previous experiments [5, 42]) or be trapped at the η′/Al interfaces by interface segregation (as revealed in Fig. 7). Combining these results in

EP

Table 2 and Fig. 7, we can further predict the solute partition of Zn, Cu, and Mg between η′ bulk and its interfaces in the alloy, as illustrated in Fig. 8. It is suggested that Cu and Mg both

AC C

strongly prefer the η′ bulk interior to its interfaces. This seemly consists well with some previous 3D-APT findings that Cu resides inside the precipitates without obvious occupation at the interfaces [5, 42, 58]. In contrast, Zn is slightly more preferred at the broad interface. The energy-favored Zn segregation to the interfacial Al1 sites on the broad interface (see Fig. 7a) agrees with the few bright spots observed on the B-layer in Fig. 6a and also with the experimental analyses in Refs. [9, 29, 58].

26

SC

RI PT

ACCEPTED MANUSCRIPT

alloy matrix.

TE D

3.3.4 Impacts on interface properties

M AN U

Fig. 8 Predicted solute partition of Zn, Cu, and Mg between η′ bulk and its interfaces in the

The high strength of Al-Zn-Mg-Cu alloys can be mainly attributed to the high number density of η′ nano-precipitates, which offers abundant η′/Al interface areas in Al matrix.

EP

Adhesion of these interfaces dictates largely the overall mechanical performance of these

AC C

alloys [59]. To assess the potential impacts of interface segregation, we further calculated the interface adhesion strength in term of the work of separation, Wsep, for both clean and segregated interfaces. Wsep was calculated as the difference in the energy between that of the equilibrium, bonded interface and that of the two fully-separated halves, divided by the cross-sectional area of the interface. The relevant energetics and the corresponding structures are summarized in Fig. 9. The stoichiometric AlZn-terminated {10-10}η′/{110}Al periphery interface has slightly higher work of separation and thus slightly higher adhesion strength

27

ACCEPTED MANUSCRIPT than the stoichiometric Al-terminated {0001}η′/{111}Al broad interface. Generally, higher adhesion allows more plasticity to be activated during fracture and thus corresponding higher interface toughness [60]. In this sense, interface dehesion may occur at the broad interface

RI PT

more easily than at the periphery interface. Also, segregated Zn has no significant impact on adhesion for both the two interfaces. Upon Zn segregation, the periphery interface adhesion is

TE D

M AN U

SC

still slightly stronger than the broad interface one.

EP

Fig. 9 Calculated work of separation for (a) the Al-terminated {0001}η’/{111}Al broad interfaces, and (b) the AlZn-terminated {10-10}η′/{110}Al periphery interface, with and

AC C

without Zn segregation

To further assess the potential impacts of interface segregation at higher Zn coverage, we evaluated the total energy reduction and the resulting work of separation as a function of the Zn coverage for the η′/Al broad interface in Fig. 10. The number of Zn atoms at the interface region varied with the Zn coverage in the interface supercell while the excess Zn atoms were put into the Al matrix supercell as substitutional solutes (not shown). By doing so, we were

28

ACCEPTED MANUSCRIPT able to keep exactly the same types and numbers of atoms among these supercells, so that a direct comparison among the total energies can be meaningful. Our calculations revealed a consistent trend for solute Zn to segregate only to the outmost layer (B-layer) of the

RI PT

Al-terminated BI, and more specifically, to substitute only the Al atoms in there. The total energy of the system is significantly reduced as increasing the Zn coverage (as seen in Fig. 10), and consequently, the final interface structure can be terminated by two almost complete

SC

Zn atom layers at the full coverage of 1 monolayer (ML). Very similar interfacial phenomena have been also observed for nano-sized precipitates in Al-Cu alloys with minor addition of Sc

M AN U

[61] or Ag [62].

Recall that solute Zn severs as a main constituting element in η′ phase and must participate actively into the precipitation of η′. The poor substitution ability of Zn in η′ bulk

TE D

interior as compared to at the broad interface, however, would lead to the rejection of excess Zn atoms to the interface during η′ growth. In this way, increasing the Zn content would induce more intensive Zn segregation to the interfaces, to further reduce the total energy of

EP

the system. At a very high potential of Zn (that is often related to a very high Zn content

AC C

especially at high temperatures), the equilibrium BI structure turns into the Zn-terminated one as predicted by the interface phase diagram (Fig. 5) or even be terminated by an almost all-Zn bi-layer (Fig. 10). In either case, the Zn-rich shell forms, retarding the inwards diffusion of Zn and other solute atoms across the interface and thus stabilizing η′ nano-disk at its low nanometer size. This scenario seems to be consistent with many experimental observations, such as the almost unchanged thickness of η′ nano-disks during its growth, and the necessity of a high Zn content to achieve high strength in an Al-Zn-Mg-Cu alloy [63, 64]. Apparently

29

ACCEPTED MANUSCRIPT from Fig. 8, solute Cu and Mg both prefer the internal sites of η′ phase and thus cannot possibly provide the same benefits as Zn does. A high Zn content and a higher aging temperature are often desired to achieve higher diffusivity and potential of Zn in Al, and thus

RI PT

a higher Zn coverage at the broad interfaces of η′. Meantime, however, this strategy may also facilitate the Zn segregation to Al grain boundaries (GBs), which greatly reduces the cohesiveness of GBs and thus the ductility of the alloy [65, 66]. Therefore, future efforts to

SC

optimize the alloy design can be focused on finding candidate alloying elements for Zn. Such alloying elements can effectively segregate to η′ interfaces as Zn does but better not to Al

M AN U

GBs, and thus can be more efficiently to stabilize η′ at its nanometer size. Under this circumstance, the Zn content can be reduced accordingly to a more intermediate level, to

AC C

EP

TE D

enable a better balance between strength and ductility of Al-Zn-Mg-Cu alloys.

Fig. 10 Total energy reduction due to Zn segregation to the broad interface as a function of the Zn coverage, and the corresponding changes in work of separation.

30

ACCEPTED MANUSCRIPT

4. Conclusions The atomic structures of η′ bulk and its interfaces in Al-Zn-Mg-Cu alloy were investigated by coupling the aberration-corrected STEM characterization and the first-

RI PT

principles calculations. Bulk substitution and interface segregation of solutes Zn, Mg, and Cu, and their partitions in the alloy matrix were thoroughly examined and discussed. Some conclusive remarks are summarized as follows.

SC

(1) A hexagonal structure with the stoichiometry of Al4Mg2Zn3 proposed by Kverneland et al is energetically more reasonable for describing η′ bulk phase, and also consistent with our

M AN U

HAADF-STEM Z-contrast images of a 7075 Al alloy.

(2) Solute Cu shows strong preference for dissolving into η′ bulk by substituting internal Zn, Mg, and Al atoms. Mg also strongly substitutes for Al in η′. Zn has only a very weak

TE D

tendency to substitute Al in η′. These solute substitution trends agree with many individual experimental observations, in support of the transformation sequence of η′⟶ηp⟶η.

EP

(3) The equilibrium interface structures of η′/Al depend sensitively on the Zn potential. The

AC C

calculated interface phase diagram suggests: within a wide range of Zn potential, the broad interface is dominated by the stoichiometric Al-terminated structure while the periphery interface is dominated by the stoichiometric AlZn-terminated structure. Only at very high Zn potentials, can the broad interface transform into the Zn-terminated structure, while only at very low Zn potentials, can the periphery interface transform into the MgZn-terminated structure. (4) Due to the large discrepancy in formation energy between the broad interface and the

31

ACCEPTED MANUSCRIPT periphery interface, η′ nano-disks tend to grow in a manner that the disk diameter increases ahead of its thickness. The interface anisotropy and hence the equilibrium aspect ratio of η′ nano-disks can be thus predicted as ~4.3.

RI PT

(5) Solute Cu and Mg both prefer to dissolve into the η′ bulk interior, while Zn is more likely to segregate to the broad interface. Zn segregation has no significant impact on interface adhesion strength until its interface coverage increases to fairly high.

SC

(6) Increasing the Zn potential (conceptually related to a higher Zn content and aging temperature) favors the formation of the Zn-terminated broad interface or an almost

M AN U

all-Zn bi-layer at the broad interface, which helps stabilize η′ at its low nanometer size by retarding the atomic diffusion of Zn and other solute species across the interface. Based on this knowledge, a new strategy can be suggested for future optimal design of

AC C

EP

TE D

Al-Zn-Mg-Cu alloys.

32

ACCEPTED MANUSCRIPT Acknowledgment The authors would like to thank the financial support from the National Science Foundation of China (No. 51471189 and 51474244) and the National Basic Research Program of China

RI PT

(sub-contract No. 2014CB644001-2) and the Fundamental Research Funds for the Central Universities of Central South University (No. 502221804). The computational resource at the High Performance Computing Center of Central South University is also gratefully

AC C

EP

TE D

M AN U

SC

acknowledged

33

ACCEPTED MANUSCRIPT References [1] X. Fan, D. Jiang, Q. Meng, Z. Lai and X. Zhang, Characterization of precipitation microstructure and properties of 7150 aluminum alloy, Mater. Sci. Eng. A 427 (2006)

RI PT

130-135 [2] P. Bai, X. Hou, X. Zhang, C. Zhao and Y. Xing, Microstructure and mechanical properties of a large billet of spray formed Al–Zn–Mg–Cu alloy with high Zn content, Mater. Sci.

SC

Eng. A 508 (2009) 23-27

[3] J. Chen, L. Zhen, S. Yang, W. Shao and S. Dai, Investigation of precipitation behavior and

M AN U

related hardening in AA 7055 aluminum alloy, Mater. Sci. Eng. A 500 (2009) 34-42 [4] L. K. Berg, J. GjØnnes, V. Hansen, X. Z. Li, M. Knutson-Wedel, G. Waterloo, D. Schryvers and L. R. Wallenberg, GP-Zones in Al-Zn-Mg alloys and their role in artificial

TE D

aging, Acta Mater. 49 (2001) 3443-3451

[5] G. Sha and A. Cerezo, Early-stage precipitation in Al–Zn–Mg–Cu alloy (7050), Acta Mater. 52 (2004) 4503-4516

EP

[6] L. F. Mondolfo, N. A. Gjostein and D. W. Levinson, Structural changes during the aging

AC C

in an Al-Mg-Zn Alloy, JOM 8 (1956) 1378-1385 [7] J. GjØnnes and C. J. SIMENSENS, An electron microscope investigation of the microstructure in an Aluminum-zinc-magnesium alloy, Acta Metall. 18 (1970) 881-890 [8] A. Kverneland, V. Hansen, R. Vincent, K. Gjonnes and J. Gjonnes, Structure analysis of embedded nano-sized particles by precession electron diffraction: η′-precipitate in an Al-Zn-Mg alloy as example, Ultramicroscopy 106 (2006) 492-502 [9] X. Xu, J. Zheng, Z. Li, R. Luo and B. Chen, Precipitation in an Al-Zn-Mg-Cu alloy during

34

ACCEPTED MANUSCRIPT isothermal aging: Atomic-scale HAADF-STEM investigation, Mater. Sci. Eng. A 691 (2017) 60-70 [10] X. Z. Li, V. Hansen, J. GJénnes and L. R. Wallenberg, HREM study and structure

RI PT

modeling of the η' phase the hardening precipitate in commercial Al-Zn-Mg alloys, Acta Mater. 47 (1999) 2651-2659

[11] H. B. Larsen, G. Thorkildsen, S. Natland and P. Pattison, Average crystal structure(s) of

SC

the embedded meta-stable η′ phase in the Al–Mg–Zn system, Philos. Mag. 94 (2014) 1719-1743

M AN U

[12] J. H. Auld, S. M. Cousland, The structure of the metastable η' phase in aluminium-zinc-magnesium Alloys, J. Aust. Inst. Met. 19 (1974) 194–199 [13] W. Lefebvre, F. Danoix, G. Da Costa, F. De Geuser, H. Hallem, A. Deschamps and M.

TE D

Dumont, 3DAP measurements of Al content in different types of precipitates in aluminum alloys, Surf. Interface Anal. 39 (2007) 206-212 [14] M. Dumont, W. Lefebvre, B. Doisneau-Cottignies and A. Deschamps, Characterization

EP

of the composition and volume fraction of η′ and η precipitates in an Al–Zn–Mg alloy by

AC C

a combination of atom probe, small-angle X-ray scattering and transmission electron microscopy, Acta Mater. 53 (2005) 2881-2892 [15] C. Wolverton, Crystal structure and stability of complex precipitate phases in Al–Cu– Mg–(Si) and Al–Zn–Mg alloys, Acta Mater. 49 (2001) 3129-3142 [16] K. Kim, B.-C. Zhou and C. Wolverton, First-principles study of crystal structure and stability of T 1 precipitates in Al-Li-Cu alloys, Acta Mater. 145 (2018) 337-346 [17] L. K. Berg, J. GJØnnes, V. Hansew, X. Z. Li, M. Knutson, G. Waterloo, D. Schryvers

35

ACCEPTED MANUSCRIPT and L. R. Wallenberg, GP-zones in Al–Zn–Mg alloys and their role in artificial aging, Acta Mater. 49 (2001) 3443-3451 [18] J.Gjønnes and C. J. Simensen, An electron microscope investigation of the

RI PT

microstructure in an aluminum-zinc-magnesium alloy, Acta Metall. 18 (1970) 881-890 [19] A. Bendo, K. Matsuda, S. Lee, K. Nishimura, N. Nunomura, H. Toda, M. Yamaguchi, T. Tsuru, K. Hirayama, K. Shimizu, H. Gao, K. Ebihara, M. Itakura, T. Yoshida and S.

Mg alloys, J. Mater. Sci. 53 (2017) 4598-4611

SC

Murakami, Atomic scale HAADF-STEM study of η′ and η1 phases in peak-aged Al–Zn–

M AN U

[20] J. Z. Liu, J. H. Chen, X. B. Yang, S. Ren, C. L. Wu, H. Y. Xu and J. Zou, Revisiting the precipitation sequence in Al–Zn–Mg-based alloys by high-resolution transmission electron microscopy, Scripta Mater. 63 (2010) 1061-1064

TE D

[21] M. Nicolas and A. Deschamps, Characterization and modeling of precipitate evolution in an Al–Zn–Mg alloy during non-isothermal heat treatments, Acta Mater. 51 (2003) 6077-6094

EP

[22] J. C. Werenskiold, A. Deschamps and Y. Bréchet, Characterization and modeling of

AC C

precipitation kinetics in an Al–Zn–Mg alloy, Mater. Sci. Eng. A 293 (2000) 267-274 [23] J. da Costa Teixeira, D. G. Cram, L. Bourgeois, T. J. Bastow, A. J. Hill and C. R. Hutchinson, On the strengthening response of aluminum alloys containing shear-resistant plate-shaped precipitates, Acta Mater. 56 (2008) 6109-6122 [24] A. Biswas, D. J. Siegel and D. N. Seidman, Simultaneous segregation at coherent and semi-coherent heterophase interfaces, Phys. Rev. Lett. 105 (2010) 076102 [25] V. Vaithyanathan, C. Wolverton and L. Q. Chen, Multiscale modeling of θ′ precipitation

36

ACCEPTED MANUSCRIPT in Al–Cu binary alloys, Acta Mater. 52 (2004) 2973-2987 [26] J. F. Nie, Effects of precipitate shape and orientation on dispersion strengthening in magnesium alloys, Scripta Mater. 48 (2003) 1009-1015

RI PT

[27] X. Z. Li, V. Hansen, J. GJénnes and L. R. Wallenberg, HREM Study and structure modeling of the η' phase, the hardening precipitates in commercial Al-Zn-Mg alloys, Acta Mater. 47 (1999) 2651-2659

SC

[28] Y. Y. Li, L. Kovarik, P. J. Phillips, Y.-F. Hsu, W.-H. Wang and M. J. Mills, High-resolution characterization of the precipitation behavior of an Al–Zn–Mg–Cu alloy,

M AN U

Philo. Mag. Lett. 92 (2012) 166-178

[29] C. D. Marioara, W. Lefebvre, S. J. Andersen and J. Friis, Atomic structure of hardening precipitates

in

an

Al–Mg–Zn–Cu

alloy

determined

by

HAADF-STEM

and

TE D

first-principles calculations: relation to η-MgZn2, J. Mater. Sci. 48 (2013) 3638-3651 [30] T. Tsuru, M. Yamaguchi, K. Ebihara, M. Itakura, Y. Shiihar, K. Matsuda, H. Toda, First-principles study of hydrogen segregation at the MgZn2 precipitate in Al-Mg-Zn

EP

alloys, Comp. Mater. Sci, 148 (2018) 301-306

AC C

[31] G. Kresse and J. Furthmüller, Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set, Phys. Rev. B 54 (1996) 11169-11186 [32] P. E. Blöchl, Projected augmented-wave method, Phys. Rev. B 50 (1994) 17953-17579 [33] G. Kresse and D. Joubert, From ultra-soft pseudo-potentials to the projector augmented wave method, Phys. Rev. B 59 (1999) 1758 [34] P. Vinet, J. H. Rose, J. Ferrante and J. R. Smith, Universal features of the equation of state of solids, J Phys: Condens. Mat. 1 (1989) 1941

37

ACCEPTED MANUSCRIPT [35] D. Ceperley and B. Alder, Ground state of the electron gas by a stochastic method, Phys. Rev. Lett. 45 (1980) 566 [36] J. P. Perdew and A. Zunger, Self-interaction correction to density-functional

RI PT

approximations for many-electron systems, Physical Review B 23 (1981) 5048 [37] G. Simmons, H. Wang, Single crystal elastic constants and calculated aggregate properties: a Handbook, M. I. T. Press, Cambridge, MA, 1971

Simple, Phys Rev Lett 77 (1996) 3865

SC

[38] J. P. Perdew, K. Burke and M. Ernzerhof, Generalized Gradient Approximation Made

M AN U

[39] S. Haussühl, Structure and Properties, Wiley‐VCH, Verlag GmbH, (1976) 55-56. [40] K. Wen, Y. Fan, G. Wang, L. Jin, X. Li, Z. Li, Y. Zhang and B. Xiong, Aging behavior and precipitate characterization of a high Zn-containing Al-Zn-Mg-Cu alloy with various

TE D

tempers, Mater. Des. 101 (2016) 16-23

[41] S. K. Maloney, K. Hono, I. J. Polmear and S. P. Ringer, The chemistry of precipitates in an aged Al-2.1Zn-1.7Mg at.% alloy, Scripta Mater. 41 (1999) 1031–1038

EP

[42] T. Marlaud, A. Deschamps, F. Bley, W. Lefebvre and B. Baroux, Evolution of

AC C

precipitate microstructures during the retrogression and re-ageing heat treatment of an Al–Zn–Mg–Cu alloy, Acta Mater. 58 (2010) 4814-4826 [43] T. Engdahl, V. Hansen, P. J. Warren and K. Stiller, Investigation of fine scale precipitates in Al–Zn–Mg alloys after various heat treatments, Mater. Sci. Eng. A 327 (2002) 59–64 [44] K. Stiller, P. J. Warren, V. Hansen, J. Angenete and J. Gjønnes, Investigation of precipitation in an Al–Zn–Mg alloy after two-step ageing treatment at 100° and 150°C,

38

ACCEPTED MANUSCRIPT Mater. Sci. Eng. A 270 (1999) 55–63 [45] J. Z. Liu, J. H. Chen, Z. R. Liu and C. L. Wu, Fine precipitation scenarios of AlZnMg(Cu) alloys revealed by advanced atomic-resolution electron microscopy study Part II: Fine

RI PT

precipitation scenarios in AlZnMg(Cu) alloys, Mater. Character. 99 (2015) 142-149 [46] J. Z. Liu, J. H. Chen, D. W. Yuan, C. L. Wu, J. Zhu and Z. Y. Cheng, Fine precipitation scenarios of AlZnMg(Cu) alloys revealed by advanced atomic-resolution electron

alloys, Mater. Character. 99 (2015) 277-286

SC

microscopy study Part I: Structure determination of the precipitates in AlZnMg(Cu)

M AN U

[47] A. Kverneland, V. Hansen, G. Thorkildsen, H. B. Larsen, P. Pattison, X. Z. Li and J. Gjønnes, Transformations and structures in the Al–Zn–Mg alloy system: A diffraction study using synchrotron radiation and electron precession, Mater. Sci. Eng. A 528 (2011)

TE D

880-887

[48] Y. Jiang, J. R. Smith and A. G. Evans, Temperature dependence of the activity of Al in diluteNi(Al)solid solutions, Phys. Rev. B 74 (2006) 224110

EP

[49] T. Engdahl, V. Hansen, P. J. Warren and K. Stiller, Investigation of fine scale

AC C

precipitates in Al–Zn–Mg alloys after various heat treatments, Mater. Sci. Eng. A 327 (2002) 59-64

[50] T. Marlaud, A. Deschamps, F. Bley, W. Lefebvre and B. Baroux, Influence of alloy composition and heat treatment on precipitate composition in Al–Zn–Mg–Cu alloys, Acta Mater. 58 (2010) 248-260 [51] A. Deschamps, A. Bigot, F. Livet, P. Auger, Y. Bréchet and D. Blavette, A comparative study of precipitate composition and volume fraction in an Al–Zn–Mg alloy using

39

ACCEPTED MANUSCRIPT tomographic atom probe and small-angle X-ray scattering, Philo. Mag. A 81 (2001) 2391-2414 [52] X. Fang, M. Song, K. Li, Y. Du, D. D. Zhao, C. Jiang and H. Zhang, Effects of Cu and

RI PT

Al on the crystal structure and composition of η(MgZn2) phase in over-aged Al–Zn–Mg– Cu alloys, J. Mater. Sci. (2012) 47 5419–5427

[53] L. F. Mondolfo, Aluminum alloys: structure and properties. Butterworths, London, 1976

SC

[54] Y. Jiang, J. R. Smith and A. G. Evans, First principles assessment of metal/oxide interface adhesion, Appl. Phys. Lett. 92 (2008) 141918

M AN U

[55] L. Yang, Y. Jiang, Y. Wu, G. R. Odette, Z. Zhou and Z. Lu, The ferrite/oxide interface and helium management in nano-structured ferritic alloys from the first principles, Acta Mater. 103 (2016) 474-482

TE D

[56] W. Zhang, J. R. Smith and A. G. Evans, The connection between ab initio calculations and interface adhesion measurements on metal/oxide systems: Ni/Al2O3 and Cu/Al2O3, Acta Mater. 50 (2002) 3803–3816

EP

[57] J. D. Robson, Modeling the overlap of nucleation, growth and coarsening during

AC C

precipitation, Acta Mater. 52 (2004) 4669-4676 [58] N. Q. Chinh, J. Lendvai, D. H. Ping and K. Hono, The effect of Cu on mechanical and precipitation properties of Al–Zn–Mg alloys, J. Alloys Compd. 378 (2004) 52-60 [59] Y. Mishin, M. Asta and J. Li, Atomistic modeling of interfaces and their impact on microstructure and properties, Acta Mater. 58 (2010) 1117-1151 [60] Y. Jiang, Y. Wei, J. R. Smith, J. W. Hutchinson and A. G. Evans, First principles based predictions of the toughness of a metal/oxide interface, Int. J. Mat. Res. 101 (2010) 1-8

40

ACCEPTED MANUSCRIPT [61] C. Yang, P. Zhang, D. Shao, R. H. Wang, L. F. Cao, J. Y. Zhang, G. Liu, B. A. Chen and J. Sun, The influence of Sc solute partitioning on the micro-alloying effect and mechanical properties of Al-Cu alloys with minor Sc addition, Acta Mater. 119 (2016)

RI PT

68-79 [62] E. Gariboldi, P. Bassani, M. Albu and F. Hofer, Presence of silver in the strengthening particles of an Al-Cu-Mg-Si-Zr-Ti-Ag alloy during severe overaging and creep, Acta

SC

Mater. 125 (2017) 50-57

[63] D. Liu, B. Xiong, F. Bian, Z. Li, X. Li, Y. Zhang, Q. Wang, G. Xie, F. Wang and H. Liu,

M AN U

Quantitative study of nanoscale precipitates in Al–Zn–Mg–Cu alloys with different compositions, Mater. Sci. Eng. A 639 (2015) 245-251

[64] Z. Chen, Y. Mo and Z. Nie, Effect of Zn Content on the Microstructure and Properties of

TE D

Super-High Strength Al-Zn-Mg-Cu Alloys, Metall. Mater. Trans. A 44 (2013) 3910-3920 [65] R. Mahjoub, K. J. Laws, N. Stanford and M. Ferry, General trends between solute segregation tendency and grain boundary character in aluminum - an ab inito study, Acta

EP

Mater. 158 (2018) 257-268

AC C

[66] S. Zhang, O. Y. Kontsevoi, A. J. Freeman and G. B. Olson, First principles investigation of zinc-induced embrittlement in an aluminum grain boundary, Acta Mater. 59 (2011) 6155-6167

41