Effect of surface grinding on the microstructure and corrosion of Al-Mg-Si-Cu alloys

Surface treatment has shown strong effects on the corrosion resistance of Al-Mg-Si-Cu alloys. To elucidate the surface grinding effects, a set of the peak-aged Al-Mg-Si-Cu samples are ground using SiC papers with different grits, and the ground surface microstructure and corrosion performance are investigated. The results show that grinding generates a deformed near-surface layer. The deformed layer is characterized by nanocrystalline and the alloying element Cu segregates at nano-grained boundaries. The Cu segregation bands increase the available cathode area and the potential difference between the grain boundary and grain interior, which enhances the microgalvanic corrosion effect between the phase α-AlFeSi (Mn, Cu) and the adjacent matrix. As the SiC paper grit number decreased from 1200 to 80, the Cu segregation bands of the deformed layers increases from 600 to 1600 nm, the corresponding corrosion current density increases from 1.0 ± 0.1 to 22.8 ± 0.7 μA.cm−2, and the corrosion potential decreases from −742 ± 13.1 to −791 ± 15.0 mV (versus SCE).


Introduction
Al-Mg-Si-Cu (6000 series) alloy sheets are widely used in manufacturing engineering products such as automotive structures, shipping containers and various domestic appliances, due to their good combination of high specific strength, high ductility, high corrosion resistance and low production cost [1][2][3][4][5].The addition of 0.6% Cu can induce the dense precipitation of Q (AlMgSiCu) phase in addition to β (Mg 2 Si), which can significantly improve the age hardening rate of the alloy by 25% [6,7].But on the other hand, the corrosion resistance of the alloys decreases rapidly with the increasing Cu content, primarily due to the Q-promoted localized (i.e.intergranular and pitting) corrosion, which can degrade the mechanical properties and even cause sudden structural failures.
Several design strategies have been developed in recent decades to address this problem, including reducing the precipitation amount of Q phase at grain boundaries [8,9], introducing other alloying elements to change the electro-potential of Q phase [10,11].Besides, some surface treatments can be also expected to have important influence, since the initial corrosion behavior is always closely related to surface conditions [12][13][14].Even a routine machining process, such as cutting, rolling or grinding, can induce a highly deformed layer on the work-piece surfaces, therefore changing the surface microstructure and microchemistry to be distinct from the underlying matrix.For instance, the Al alloy surface could undergo severe shear deformation during the rolling, resulting in a nano-size fine grain structure [13,15].Such a fine grained surface more easily suffer from intergranular corrosion that tends to initiate at the surface grain boundaries, as manifested in various 3000, 5000 and 6000 series Al alloys [8,16].Liu et al [8] further established a close correlation of Q phase and cosmetic (filiform) corrosion in an AA6111 alloy, where matrix-precipitating and GB-precipitating Q phase can induce a larger electro-potential difference of Q and GB with the matrix, respectively, thereby expediting both pitting and intergranular corrosion.Similar findings were also suggested for other intermetallic particles (IMP) in some 3000 and 5000 series Al alloys [16].Moreover, several independent studies observed the fragmentation and dissolution of IMPs during surface grinding, such as η phase (MgZn 2 ) in 7000 series Al alloys [17,18], leading to the redistribution of solute elements in the near-surface regions that may have more intricate influence on the corrosion performance.But nevertheless, there is still no evidence that such solute re-dissolution and redistribution can also occur to Q phase in Al-Mg-Si-Cu alloys.
Furthermore, given an alloy surface, different surface treatments would end up with different levels of surface roughness that can also affect the alloy corrosion performance.SAJJAD et al [19] noticed that larger and deeper machining-induced surface grooves always correspond to the enhanced corrosion kinetics in an AA5050 alloy, as similarly found in many other alloys (such as a 2A97 alloy by NIU et al [20], an AA 6061 alloy by MAURYA et al [21], and an AA7075 alloy by FRANKEL et al [14].These findings are generally expected, for that a severer deformation imposes more comprehensive influence on the near-surface microstructure, in addition to an increased surface area.
As mentioned above, the alloying with Cu can largely improve the mechanical properties but degrade the corrosion resistance of Al-Mg-Si-Cu alloys.The corrosion resistance can be further greatly affected by surface finishing.This present work was devoted to elucidate the effects of surface grinding on the corrosion performance of Al-Mg-Si-Cu alloys.To meet this goal, several ground samples of an AA6111 alloy were prepared using SiC sandpapers with different grits.The resulting surface topographies, grain and precipitation microstructures, and electrochemical performances were thoroughly characterized.The dissolution of Q phase during surface grinding was confirmed in the AA6111 alloy.Only the dissolved Cu atoms can strongly segregate to Al GBs, which was believed to change the electro-potentials of the GBs and hence bring inverse influence on alloy corrosion performance.The newly gained insights can not only improve our understanding of surface deformation and its relation with alloy corrosion performance, but also guide us for the optimal design of new Al-Mg-Si based alloys.

Material and heat treatment
The as-cast AA6111 alloys have the nominal chemical composition of Al-0.8Mg-1.0Si-0.3Mn-0.9Cu-0.27Fe-0.13Zn-0.11Ti(all in wt%, hereafter).The alloy ingots were produced by melting master alloys Al-50Cu, Al-10Si and Al-10Mn with pure Mg, Zn, Fe and Ti in a resistance furnace and casting into a water-cooled iron mold.Table 1 lists the real compositions of the ingots measured by ICP-OES (the inductively coupled plasma optical emission spectrometry).The ingots were then homogenized at 550 °C for 30 h, and then hot and cold-rolled to a final sheet of 2 mm in thickness.The alloy samples were cur from the sheet and then solution treated at 600 °C for 40 min followed by water quenching.The quenched samples were immediately artificially aged at 180 °C for 8 h (T6).The aged samples were then cut into thin pieces of 10 × 10 × 2 mm (the type of wire-cutting machine used is DK7735.Table travel: 350 mm × 450 mm, cutting thickness: 380 mm) and divided into two groups: one group were ground in ethyl alcohol using the SiC sandpapers of 80, 400 and 1200 grit, denoted as 80-grit, 400grit and 1200-grit, respectively, for electrochemical property tests; The other group were ground with the SiC sandpaper of 1500 grit and polished in a suspension of colloidal silica (0.06 μm).The polished samples were then cleaned using deionized water and air-dried, for atomic force microscope (AFM) observation.

Microstructure characterization
Surface features of the 80, 400 and 1200-grit samples were examined using an optical microscope (OM, OLYMPUS DSX500).Surface roughness (R a ) was measured using a laser scanning confocal microscope (ZEISS Axio LSM700).Microstructure observations were performed using a scanning electron microscope (SEM, ZEISS EVOM10) equipped with an energy dispersive x-ray spectroscopy (EDS, One Max 50).The effective grain size were measured by Image J software.Nano-precipitates and grain boundaries were characterized using a scanning transmission microscope (S/TEM, Talos F200X) at 200 kV and a high resolution transmission electron microscopes (HRTEM, FEI Titan G2 60-300) at 300 kV.High-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) with Z-contrast imaging employed a Thermo Fisher Talos F200X.The spot brightness depends on the atomic number (Z) and the number of atoms along the atomic column.TEM specimens were prepared by twin-jet electro-polishing in a solution of 25 vol% nitric acid in methanol at 18 V under −30 °C.Focused ion beam (FIB, FEI Helios Nanolab 600i) was used to cut the cross-sectional specimens of the near-surface regions for TEM observations.

Electrochemical and corrosion tests
Scanning Kelvin probe force microscope (SKPFM) was used to determine surface topographies and the Volta potentials of alloy surfaces, using a Si probe tip with a CoCr coating.The measurements were fulfilled in an atmospheric environment at a scanning rate of 0.5 Hz.The AFM model used is: BRUKER Dimension FastScan, the maximum scanning size of the FastScan probe: 33 * 33 μm 2 , Z direction: 4 μm; Scan speed up to 6 Hz-20 Hz.
The measured results were analyzed using NanoScope Analysis 1.8 software.Electrochemical property tests adopted a three-electrode system, using the ground alloy sample pieces of 10 × 10 × 2 mm as the working electrode, a platinum sheet as the auxiliary electrode, and a saturated calomel electrode (SCE) as the reference electrode.The test solution was selected to be 3.5% NaCl.A scanning rate of 0.33 mV s −1 was used for the polarization curve test.Corrosion potential (E corr ) and corrosion current (I corr ) were obtained by Tafel curve fitting.Electrochemical impedance spectroscopy (EIS) measurements were performed using a 10 mV AC signal with the frequency range of 0.01 ∼ 100000 Hz at the open circuit potential (OCP) of each specimen.The EIS results were fitted with a Randles-type equivalent circuit with ZSimpWin software.Each data was averaged on three repetitive tests.Immersion corrosion tests were conducted in 3.5% NaCl solution at 25°C for 24 h, and during immersion, the samples were placed flat.After the immersion tests, the samples were cleansed in HNO 3 solution at 25 °C for 5 min to remove corrosion products.The corroded surface morphologies were observed by OM, SEM and optical surface profilometer (Wyko NT9100).

Surface topography
Surface morphologies and roughness are compared for 80-grit, 400-grit and 1200-grit samples in figure 1.The sample ground with SiC sandpaper shows many noteworthy deep scratches along the grinding direction, which are regularly arranged on the sample surface (see figures 1(a)-(c)).It is noted that, wear traces of samples ground with 80-grit specimen are relatively deeper and narrower than that of the milled one with 1200-grit.
Grooves on the surface are produced by the cutting of abrasive particles, which are closely related to the grit size (grit number).The surface roughness of the samples varies significantly with different grit numbers of the SiC sandpaper employed.After grinding with 80-grit number sandpaper, obvious ridge and parallel valley appeared along the scratch direction, and the average depth and width are 56.15 and 25.93 μm, respectively.As the grit number of sandpaper increases from 80 to 1200, the wear marks become shallower, the sample surfaces become smoother (as shown in figures 1(g)-(i)), and the surface roughness (R a ) decreases from 18.03 to 6.12 μm.

Coarse intermetallics
Figure 2(a) shows the inclusion phases with as-cast Al-Mg-Si-Cu alloy, which cannot be dissolved in subsequent heat treatment.Combined with the element mappings and EDS spectral analysis, it can be inferred that the white particles are α-AlFeSi (Mn, Cu) phase (blue box in figure 2(a) with red arrows).The average size of the α-phase is 1.98 μm, and their average chemical composition is 89.68 wt% Al, 3.26 wt% Si, 1.98 wt% Mn, 2.73 wt% Fe, and 2.45 wt% Cu.However, such inclusion phase is broken in the rolling process and has a banding distribution.
SKPFM technology is extensively applied to study the localized corrosion behavior of aluminium alloys [22][23][24][25].The corrosion behavior of phases in the Al-Mg-Si-Cu alloy can be further analyzed by measuring surface volta potential.The typical test results in figures 2(b), (c) show the AFM morphology and volta potential distribution region in the blue box in figure 2(a).As seen from the volta potential distribution, α-phase shows higher volta potential than the surrounding matrix.The changes in the voltage potential at lines 1-3 in figure 2(d) show significant difference between a-phase and the matrix.According to the above results, there is a potential difference between the a-phase and the adjacent matrix.For example, α-AlFeSi (MnCu), which is mainly composed of noble elements such as Cu, Fe, etc, has a higher potential than that of the surrounding matrix, and its chemical behaviour is significantly different from that of the surrounding matrix phase.When the αphase forms galvanic coupling with aluminium matrix, the α-AlFeSi(MnCu) acting as the cathode which provides a higher rate of redox reaction [26,27], and this results in a preferential dissolution of the surrounding aluminium matrix.Thus the chemical behavior of α-phase is usually significantly different from that of the surrounding matrix phase, and the potential difference is the driving force of pitting corrosion.Pitting corrosion has a large latent disease, which causes intergranular corrosion, spalling corrosion, etc, and brings a great threat to the safe use of equipment.

Near-surface microstructure of ground samples
Figure 3 shows the bright field TEM images of the ground alloy samples.Beneath the Pt protective layer is the gradient microstructure developed in each sample, showing a microstructure transition from deformationinduced ultra-fine grained surface region to a high density dislocation zone, further to the un-deformed bulk interior.The surface region has a thickness of 1600, 1000 or 600 nm in the 80, 400, or 1200-grit sample, respectively, being separated from the un-deformed bulk interior by a high density of dislocation complexes (as marked by A in figures 3(a)-(c)).The deformed surface region is featured with an ultra-fine grain structure mixing with equiaxed and elongated nano-grains in figure 3(a).The electron diffraction pattern corresponding to the blue box region is also provided as the insert in figure 3(a), showing a typical polycrystalline ring arising by a random orientation distribution of ultra-fine grains.The widths and lengths of elongated grains ranged from 60 ∼ 150 nm and 200 ∼ 400 nm, respectively.The diameters for equiaxed grains were measured as 70 ∼ 100 nm.It is clearly seen from the higher magnification bright field image in figure 3(d) that partial nano-grains contain a very high dislocation density while the rest contain a much lower dislocation density, showing a combinational result of dynamics recrystallization and severe plastic deformation induced by surface deformation.The similar gradient microstructure characteristics with a fine-grained surface region has been also reported for the rolled surfaces of AA3005 and AA5754 alloys in [16].Clearly, the GB is enriched with Cu atoms.In figures 4(c) and (d), compared with the uniform distribution of the bulk alloy hardening precipitates Q′, the strengthening phases Q′ are largely absent in the near-surface layer.This suggests that the precipitate nano-phase is thermodynamically unstable and eventually dissolves [28], due to the sharp increase in surface free energy.EDX analysis reveals the enhancement of Al, Mg, Si, Cu at one nano-particle  mappings and EDS spectral analysis, these precipitates could be identified as Q′ phase [29,30].Q' phase is hexagonal, with its habit plane of {510}, a unit cell a = 1.032 nm, c = 0.405 nm.

Electrochemical performance
The open circuit potential (OCP) of the Al-Mg-Si-Cu aluminum alloy with three surface finishes in 3.5% NaCl solution is measured.As shown in figure 6(a), the OCP curves gradually increase and then level off.The potential of the sample reaches a stable state in a short time due to the formation of an oxide film in contact with the air of the sample.The potential of the stable state decreases as the grit number increases.The OCP of the 80-grit, 400grit and 1200-grit samples increases gradually from −795 to −783 mV (SCE), −761 to −751 mV (SCE) and −753 to −746 mV (SCE), respectively.The OCP initial potential and stable potential of the rough 80-grit sample are −795 mV (SCE) and −783 mV (SCE) , respectively.Compared to the rough 1200-grit sample, the initial potential (−753 mV (SCE)) and the stable potential (−746 mV (SCE)) of the 1200-grit smooth sample move to positive direction.
This suggests that the smoother surface has a lower activity.Figure 6(b) shows potentiodynamic polarization curves of different test samples.Based on Tafel extrapolation [31,32], the corrosion potential (E corr ) and the corrosion current density (j corr ) obtained from the potentiodynamic polarization curves of the Al-Mg-Si-Cu alloy with three different surface finishes are listed in table 2. The corrosion modes of these samples are similar.Corrosion potential and corrosion current are usually employed to evaluate the corrosion resistance of aluminum alloys.The corrosion current is a kinetic parameter.The greater the value is, the higher the corrosion rate is [33].With the increase of the grit number of sandpaper, E corr of samples gradually moves to the positive direction.E corr of the surface ground with 80-grit sample is −791 mV (SCE), whereas the value of the 1200-grit sample is −742 mV (SCE).According to Faraday's Second Law, the corrosion current density (j corr ) is positively related to the corrosion rate.From 80-grit to 1200-grit, the j corr of surface ground sample decreases by 21.8 μA.cm −2 .The contact area between the surface and the corrosive solution increases as the abrasive roughness increases.This results in an increase in the reaction area, a decrease in the corresponding measured reaction resistance, and an increase in current.The impedance spectra and equivalent circuit of the test samples are shown in figures 6(c), (d).The Nyquist curves of three ground samples are composed of low-frequency capacitive reactance arc and medium-high frequency capacitive reactance arc.The samples exhibit similar impedance characteristics, implying that they have similar corrosion processes.Among the tested samples, the capacitive arc radii of 80-grit, before grinding, 400-grit and 1200-grit samples increase successively, and radius value of the 1200-grit sample is the largest.To quantitatively evaluate the corrosion of the Al-Mg-Si-Cu alloy with different surface ground samples, the electrochemical parameters of each sample are calculated by a ZSimpWin software.Which shows that from 80-grit to 1200-grit ground surface, R t increases by 129.5 Ω.cm 2 .And a higher R t value leads to a better corrosion resistance [34,35].

Corrosion morphology
Figures 7 and 8 present the surface corrosion morphology of different samples of the Al-Mg-Si-Cu alloy after immersion for 24 h in 3.5% NaCl solution.Figures 7(b) and (c) is secondary electrons and corresponding backscatter images, respectively.It can be seen that there are dense black corrosion spots and some larger pits.This is because the surface is rougher before grinding and there are many points of corrosion initiation.As shown in figure 8, for all ground samples, it can be seen that pitting corrosion appears on each surface, but no surface dissolution is observed.The sample ground by 80-grit sandpaper appears many dark and thick corrosion products (see figure 8(a)).Correspondingly obvious pitting corrosion is observed on the surface of the sample (figures 8(d), (g)), By contrast, the dark corrosion products from the ground surface of 400-grit sandpaper are less than those from the ground surface of the 80-grit sandpaper (figure 8(b)), it can be seen that the level of pitting on the surface is also relatively small (see figures 8(e), (h)).It can also be seen that few dark corrosion products appear on the surface of 1200-grit sample, and slight pitting can be observed on the surface (see    From 80-grit to 1200-grit grinding, the average size of corrosion pits decreases in turn.This is closely related to the near-surface deformed layer, as the grit number increases, the thickness of the fine deformed layer decreases, and the current density of the surface relatively reduces, so the pitting propagation rate of the smooth surface is slower than that of the rough surface.Compared with the rough surface, the smooth surface has better pitting resistance.Therefore, the pit size of the smooth surface tends to be smaller.According to the results of element maps (see figure 8(j)), it can be inferred that the particle in the corrosion pit is α-AlFeSi (Mn, Cu) phase.As shown in figure 9, the two-dimensional profile of representative pits after removal of corrosion products can directly describe the evolution of surface morphology.All three samples exhibit localized corrosion, with deeper and larger cavities observed on rougher surfaces.As grit number increases from 80 to 1200, the corresponding average depth of pits reaches about 18, 10, 6 μm, respectively.It can be seen that the growth rate of pits along the parallel direction is faster than that along the vertical direction [36].From 1200-grit to 80-grit grinding, the thickness of the surface deformed layer increases, and Cu segregates at nano-grained boundaries of the near-surface deformed layer, enhancing the galvanic coupling corrosion effect of α-AlFeSi (Mn, Cu) and the adjacent matrix.On rougher surfaces, pits are easy to initiate and spread; On the other hand, the pits on smoother surface are difficult to initiate and the propagation rate is lower.

Pitting mechanism
Corrosion behaviors of aluminum alloys are closely associated with IMPs, grain boundary precipitation, grain size and precipitate free zone.The electrochemical inhomogeneity of metallic material caused by intermetallic phases leads to the preferential dissolution of IMPs and/or adjacent matrix [37][38][39].Initially, corrosion occurs at the site of defect on the surface, such as intermetallic particle.α-AlFeSi (Mn, Cu) particle is the intermetallic particle of Al-Mg-Si-Cu aluminium alloys.As a cathode phase, it will cause surface corrosion around the adjacent matrix due to the potential difference between the intermetallic particle and the matrix.In addition, there are plenty of corrosive ions in the corrosion experiments, for example Cl − .The pitting corrosion process of Al-Mg-Si-Cu alloy is shown in figure 10.The initial corrosion process, aggressive Cl − ions penetrate into the matrix through defect sites arising from α-AlFeSi (Mn, Cu) particles, Cl − will dissolve the oxide film on the metal surface in this process, participating in the localized dissolution of the alloy interface [38].The galvanic corrosion effect between the α-AlFeSi (Mn, Cu) phase and the adjacent matrix causes pitting.In the deformed near-surface layer, Cu is uniformly enriched at the nano-grained boundary, which promotes the formation of new effective cathode sites.As an effective external cathode, the Cu-rich zone plays an important role in promoting the continuous propagation of pitting corrosion [11].Because of the higher corrosion rate of the deformed layer by grinding surface, these cavities tend to develop in parallel directions [40,41].As shown in figure 9, these cavities of the three samples grow faster in the parallel direction than that in the vertical direction [36].Simultaneously, pits continue to expand due to their autocatalytic properties [42].

Quantitative relationship between corrosion level and morphology
In the grinding process, strong plastic deformation occurs at the near-surface, ultrafine grains occur, and nanograined gradient microstructure appears.This phenomenon is similar to other high shear strain processes [8,13].The nanophases (Q΄ phases) are sheared and dissolved in the near-surface deformed layer.The Cu-rich bands are formed by the segregation of dissolved Cu at the nano-grained boundary during grinding in deformed layer (see figures 4(a), (b) and 5(b)).In the initial stage of corrosion, the α-AlFeSi (Mn, Cu) phases provide both the initiation point and the current needed to corrode the matrix around microgalvanic corrosion [11].At this time, the Cu segregation bands increase the available cathode area and the potential difference between the grain boundary and grain, strengthening the galvanic corrosion effect between α-AlFeSi (Mn, Cu) and the adjacent matrix [11].As shown in figure 11, with decreasing the SiC paper grit number from 1200 to 80, the thickness of the nano-grained layers increases from 600 to 1600 nm, more Cu segregates at nano-grained boundaries and dislocations, the corresponding corrosion current density increases from 1.0 to 22.8 μA.cm −2 , and the corrosion potential decreases from −742 to −791 mV (versus SCE).Therefore, for rougher surface, Cl − is more likely to penetrate into the matrix through α-AlFeSi (Mn, Cu) phase, and increase the microgalvanic corrosion effect between α-AlFeSi (Mn, Cu) and the near-surface layer.As shown in figure 12, the total number of pits produced by 80, 400 and 1200-grit samples are 2232, 1378 and 639 respectively.The average pits areas are 3205, 2151 and 929 μm 2 respectively.Compared with the rougher surface of the specimen, the corrosion current density of the smoother surface with thinner deformed layer is lower.As a result, corrosion is more difficult to initiate on the smoother surface, and the number of pits is less.From 80-grit to 1200-grit grinding, the average pits size decreases gradually, and the corresponding total pits area decreases gradually.Therefore, the propagation rate of pits is slower on the smoother surface.And it shows that the 80-grit ground surface (see figure 8) is severely pitted, however, slight pitting (see figure 8) appears on the 1200-grit ground surface.Thus, smoother surface produces lower corrosion reactivity and dissolution rate and thus exhibit better corrosion resistance.
Alloys can be induced with different surface roughness during production, and the surface roughness determines the groove opening size and the real surface areas [43].For 80-grit, 400-grit, and 1200-grit ground samples, 25 surface grooves are measured for each sample with statistical data as shown in table 3. The smaller w/d value indicates that the groove is relatively deeper and narrower.where w expresses the peak-to-peak width of the groove, d represents the height of peak and valley depth.As the surface roughness of the samples increases,   the values of w/d become smaller.The surface curve of the sample is more undulating and the wave peak is much sharper.The hydrolysis of metal cations are difficult to diffuse outward from pits, leading to more severe acidification [44,45].By changing the surface condition, the open degree of surface grooves has obvious influence on localized corrosion of stainless steel, which is usually under diffusion control condition [46,47].For this study, no diffusive impedance is observed in the EIS analysis.The results indicate that surface corrosion of the Al-Mg-Si-Cu alloy is under activation control but not diffusion control [48].Hence, the surface grooves of the Al-Mg-Si-Cu alloy have little effect on surface corrosion.Therefore, grinding introduces a near-surface deformed layer shows significant influence on pitting corrosion of Al-Mg-Si-Cu aluimium alloy.

Conclusions
(1) Grinding introduces a near-surface deformed layer of Al-Mg-Si-Cu aluimium alloy.The near-surface deformed layer contains ultrafine grains and high-density dislocations.The alloying element Cu segregates at nano-grained boundaries, and the strengthening phase Q′ is initially largely absent within the nearsurface deformed layer.From 1200-grit to 80-grit grinding, the thickness of near-surface deformed layer increases from 600 to 1600 nm.
(2) The Cu segregation bands increase the available cathode area and the potential difference between the grain boundary and grain, which strengthen the galvanic corrosion effect between α-AlFeSi (Mn, Cu) and the adjacent matrix.
(3) The corrosion current density increases with the thickness of deformed layer.As the grit number increases from 1200 to 80, the corrosion current density increases from 1.0 to 22.8 μA.cm −2 , pits are easy to initiate and spread.
(4) This research is helpful to further improve our understanding of surface deformation and its relation with alloy corrosion performance, and guide us for the optimal design of new Al-Mg-Si based alloys.

Figures 4 (
a), (b) shows the HAADF micrographs of the deformed surface region in figure 3(a) and the corresponding element mapping of Cu.The bright field and the dark field images of the high strain deformed region and bulk alloy are shown in figures 4(c) and (d), respectively.Figure 5(a) shows the HAADF micrographs of the magnified cross-section of the near-surface region marked using a blue box in figure 3(a).
Figure 5(b) shows the EDS elemental profiles near a grain boundary along the yellow line in figure 5(a).

Figure 2 .
Figure 2. The polished Al-Mg-Si-Cu-T 6 alloy: (a) the element maps of Al, Si, Mn, Fe and Cu of the region in the blue box, (b) The AFM surface topography, (c) the volta potential distribution of region in (a), and (d) the volta potential variation along the lines 1-3.

Figure 3 .
Figure 3. (a)-(c) Bright field TEM images of the ultramicrotomed cross-sections of deformed surface regions in the 80, 400, and 1200grit ground samples, respectively.A zone: high-density dislocation complex.(d) Bright-field TEM images showing the ultra-fine grain structure in the blue box in (a).

Figure 4 .
Figure 4. (a) HAADF micrograph of the deformed surface region in figures 3(a) and (b) the corresponding element mapping of Cu. (c), (d) are the bright field and the dark field images of the high strain deformed region and bulk alloy respectively.

Figure 5 .
Figure 5. (a) HAADF micrograph of the deformed surface region in figures 3(a) and (b) EDX line scan along the yellow line in (a).(c) HAADF micrograph of the un-deformed bulk interior in figures 3(a) and (d) EDX analysis of one Q΄ precipitate in (c).(e) HRTEM image and (f) the corresponding FFT pattern of Q΄ phase.

Figure 6 .
Figure 6.(a)-(d) OCP curves of the Al-Mg-Si-Cu alloy with three surface finishes in 3.5% NaCl, Potentiodynamic polarization curves three different surface finishes, Nyquist plots and equivalent circuit used in EIS analysis, respectively.

Figure 7 .
Figure 7. Surface corrosion morphology of Al-Mg-Si-Cu after 24 h immersion in 3.5% NaCl solution (before gringing).(a) OM photos; (b) the corresponding Secondary electron SEM image; (c) the corresponding Backscattered electron SEM images.

Table 2 .
figures 8(c), (f)).From 80-grit to 1200-grit grinding, the average size of corrosion pits decreases in turn.This is closely related to the near-surface deformed layer, as the grit number increases, the thickness of the fine deformed layer decreases, and the current density of the surface relatively reduces, so the pitting propagation rate of the smooth surface is slower than that of the rough surface.Compared with the rough surface, the smooth surface has better pitting resistance.Therefore, the pit size of the smooth surface tends to be smaller.According to the results of element maps (see figure 8(j)), it can be inferred that the particle in the corrosion pit is α-AlFeSi (Mn, Cu) phase.As shown in figure9, the two-dimensional profile of representative pits after removal of corrosion products can directly describe the evolution of surface morphology.All three samples exhibit localized corrosion, with deeper and larger cavities observed on rougher surfaces.As grit number increases from 80 to 1200, the corresponding average depth of pits reaches about 18, 10, 6 μm, respectively.It can be seen that the growth rate of pits along the parallel direction is faster than that along the vertical direction[36].From 1200-grit to 80-grit grinding, the thickness of the surface deformed layer increases, and Cu segregates at nano-grained

Figure 8 .
Figure 8. Surface corrosion morphology of Al-Mg-Si-Cu after 24 h immersion in 3.5% NaCl solution.(a), (b), (c) for OM photos, ground with 80-grit, 400-grit and 1200-grit respectively; (d), (e), (f) for the corresponding Secondary electron SEM image.(g), (h), (i) for the corresponding Backscattered electron SEM images; (j) the element maps of Al, Si, Mn, Fe and Cu of the region in the blue box in figure 8(i).

Figure 10 .
Figure 10.Schematic of pitting corrosion of the near-surface deformed layer in Al-Mg-Si-Cu-T6.

Figure 11 .
Figure 11.(a) The thickness of nano-grained layer varies with different grits; (b) variation the corrosion current density (j corr ) with thickness of nano-grained layer.

Table 1 .
Measured chemical compositions of the alloys (all in wt%).