Microstructure of a Heat Treated Nickel-Aluminum Bronze and Its Corrosion Behavior in Simulated Fresh and Sea Water

Nickel-aluminum bronzes (NAB) are a family of alloys with excellent mechanical and corrosion resistance properties, even in highly corrosive environments, attributed to the complex as-cast microstructures. Heat treatment, in-service conditions or welding procedures induce microstructural transformations resulting in changes of the properties. To evaluate the corrosion behavior of one selected NAB alloy, different heat treatments were performed and the relation between microstructural changes and corrosion behavior was studied. Under common casting conditions, CuAl10Fe5Ni5 bronze consists of the a-phase, b-phase, and several intermetallic phases, collectively referred to as j-phases. The microstructures of the as-cast and heat treated samples were characterized by metallographic methods. After grinding and polishing, the samples were etched with FeCl3-as well as Cu(NH4)2Cl4-solutions and examined by light optical and scanning electron microscopy. Finally, electrochemical corrosion tests of both as-cast and heat treated samples were carried out in simulated fresh water and sea water under potentiostatic conditions. Four specimens of each sample type were polarized simultaneously at different potentials for 70 h. These specimens were used to study the corrosion progress of the different specific microstructural features and constituents by metallographic investigations. Manuscript received March 22, 2016; accepted for publication June 1, 2016; published online August 11, 2016. 1 Technische Universität Wien, Inst. of Chemical Technologies and Analytics, Vienna, A-1060, Austria 2 Technische Universität Wien, Inst. of Chemical Technologies and Analytics, Vienna, A-1060, Austria (Corresponding author), e-mail: roland.haubner@tuwien.ac.at 3 Univ. of Cantabria, Dept. of Materials Science and Engineering, Santander, 39004 Spain Copyright VC 2016 by ASTM International, 100 Barr Harbor Drive, P.O. Box C700, West Conshohocken, PA 19428-2959 689 Materials Performance and Characterization doi:10.1520/MPC20160029 / Vol. 5 / No. 5 / 2016 / available online at www.astm.org


Introduction
Nickel-aluminum bronze (NAB) is a family of alloys widely used to produce cast components for marine applications owing to its excellent corrosion resistance in sea water.The microstructure of the alloy is determined by the cooling process during casting and may change by heat input, e.g., from welding, grinding or other sources, resulting in phase transformations.
During solidification of a bronze melt with 9 wt.% Al, a copper solid solution, referred to as b-phase, is formed, and in this b-phase, the other alloying elements are dissolved.Cooling below 940 C (1724 F) causes the b-phase to decompose into a-phase and j-phases.Between 940 and 800 C (1724 and 1472 F), a ternary phase region with a-, b-, and j-phases exists, and below 800 C (1472 F) only a-and j-phases are present in NAB microstructure (Fig. 1).Particularly j-phases show different morphologies and variable compositions between Fe 3 Al and NiAl [5][6][7].During quenching from 1000 C (1832 F) to room temperature, the formation of a martensitic phase, referred to as b 0 , is described.After a subsequent heat treatment, the b 0 -phase decomposes into fine-grained a-and j-phases [7].
The corrosion behavior of NAB in sea water has been repeatedly studied [8][9][10][11][12], mostly related to as-cast material.Its excellent resistance to high chloride environment is based on formation of a passivating layer, dominated by oxides of copper and aluminum.The multiphase structure bears the risk of dealloying or selective corrosion of the most reactive (anodic) phases.In particular, crevice conditions create a high risk of phase selective corrosion along lamellar j III -phase [13].Conversely, corrosion of NAB in fresh water does not appear to be much discussed in literature; apparently it is not of great technical relevance.Nevertheless, a specific case of NAB corrosion in river water has drawn our attention to this topic [14].
Based on these considerations, the influence of heat treatment on the NAB's microstructure and its influence on the corrosion behavior in fresh and seawater are the objectives of this investigation.

Experimental Procedures
An ingot of NAB CuAl10Fe5Ni5-C (according to EN-1982 [15]) was used for the investigations.The chemical composition analyzed by X-ray fluorescence (XRF) is provided in Table 1.The ingot was cut and plates of 50 by 25 by 20 mm 3 in size were used for heat treatment experiments.In this work, the material in the as-cast state was designated AC, see Table 2.   HEAT TREATMENT OF THE BRONZE Various heat treatments were carried out in a muffle-type furnace.To prevent oxidation, the NAB samples were embedded in graphite.The first heat treatment was performed at 950 C (1742 F) for 1 h.The specimens were cooled using different cooling rates: one sample was cooled in air (sample T1) and others were water quenched (samples T0, T2, T3).The second heat treatment was performed at 700 or 500 C (1292 or 932 F) for 1 day after water quenching (samples T2, T3).Table 2 summarizes the parameters for all types of samples.Finally, small sheets, of approximately 20 by 20 by 3 mm 3 in size, were cut and ground with 600 grit SiC for electrochemical corrosion tests.

METALLOGRAPHIC PREPARATION
The various samples were sectioned with a cut-off machine and warm embedded in phenolic resin.Preparation started with 320 grit SiC for plane grinding, followed by fine grinding and polishing with 9, 3, 1 lm diamond suspensions and finally with 0.05 lm alumina suspension.To reveal the microstructures, FeCl 3 -and cupric ammonium chloride (NH 4 ) 2 CuCl 4 Á2H 2 O-solutions were used [16].Metallographic investigations were performed with a light optical microscope (LOM) and a FEI QUANTA 200 K scanning electron microscope (SEM), equipped with energy dispersive X-ray analysis (EDX).Images were taken in back scattered electron (BSE) mode.
Vicker's hardness measurements HV5 were carried out to evaluate the changes in mechanical properties caused by the heat treatments (Table 2).

ELECTROCHEMICAL CORROSION TESTS
Simulated sea water (SSW) and simulated fresh water (SFW) were used as electrolytes in the tests.SSW was prepared according to the standard DIN 50,905-4, section 4.1.by dissolving 1 mmol NaHCO 3 and 0.5 mmol Na 2 SO 4 per liter distilled water, SFW was prepared, which is characterized by chloride free, buffered (60 mg/L bicarbonate), and sulfate dominated (48 mg/L).This composition was chosen on the basis of experiences from a failure analysis reported previously [14].
The saturated calomel electrode (SCE, 0 mV SCE ¼ þ242 mV H ) was used as reference electrode in all tests and all potentials is referred to SCE.
Potentiostatic measurements were carried out in a cylindrical container (ca.300 mm diameter) serving as test cell and filled with 6 L of electrolyte.A cylinder made from stainless steel mesh was placed inside along the container wall, serving as counter electrode.The reference electrode was inserted from top in the center of the container.For each test, four specimens, serving as working electrodes, were contacted by clamping to suitably bent stainless steel wires and placed symmetrically around the reference electrode.The four specimens were polarized to þ52, þ150, þ263, and þ368 mV SCE , respectively, in SFW.In SSW, the four specimens were polarized to À262, À211, À156, and À104 mV SCE .The lowest of these values was estimated being close to the free corrosion potential of the NAB in the corresponding electrolyte.Polarization was controlled by a multichannel potentiostat (Octopoti, home made [17]), hooked to a data acquisition system logging the individual currents up to > 70 h.

METALLOGRAPHY OF THE HEAT TREATED NAB ALLOY
For the as-cast NAB, Sample AC, the detailed manufacture parameters, e.g., cooling rate, are unknown.The microstructure is determined by melt solidification and solid state transformations, thus highly governed by the crystallization of a-and b-phase from the melt, while the alloying elements tend to segregate according to their diffusion coefficients (Fig. 1).The a-phase contains more Ni but less Al and Fe, contrary b contains more Al and Fe but less Ni.Below 940 C (1724 F) the b-phase starts to decompose into a and j phases and at room temperature only these phases exist.
First, coarse, spheroidal j I precipitates (Fe 3 Al) are formed, followed by the finer j II (Fe 3 Al) and the lamellar j III -phase (NiAl).It is obvious that in areas with higher concentrations of Al, Fe, and Ni, the amount of j-phases is increased.At last, small j IV particles are precipitated (Fe 3 Al).
In Fig. 2, LOM and SEM-BSE images are shown, and by etching, the different phases are revealed.After etching with FeCl 3 , the microstructure is well contrasted and the Fe 3 Al-based j-phases (k I , k II , k IV ) are attacked more severely (Fig. 2a).Etching with Cu(NH 4 ) 2 Cl 4 colors the a and j III yellow and brown, in contrast j I , j II , and j IV are bluish (Fig. 2b).In SEM-BSE images, the j-phases are dark and the Cu-rich a-phase is bright (Fig. 2c).The average Vicker's hardness is 160 HV5.
First Heat Treatment at 950 C (1742 F) for 1 h and Cooling NAB sample T0 was heat treated at 950 C (1742 F) and quenched in water (fast cooling).According to the equilibrium diagram in Fig. 1, at 950 C (1742 F) a-and b-phase exist, and the other alloying elements are dissolved.However, the microstructures of sample T0 in Figs. 3 reveal that j I ,and j II -phases were undissolved (Fig. 3a).Explanations for this are the short solution annealing time or deviation from the equilibrium diagram in Fig. 1.During quenching, the diffusion of the alloying elements is suppressed and b is transformed to b 0 , referred to as martensite (Fig. 3c).In the microstructure of T0 the coarse plate-like a-phase is embedded in a fine-needled b 0 -phase.j I , j II (bluish, Cu(NH 4 ) 2 Cl 4 etchant) are visible in a-and b 0phase, j III is present in fine-needled b 0 -phase only (Fig. 3b).The j IV precipitates are not observed in the a-phase any more.The average Vickers hardness is 232 HV5.
Sample T1 was heat treated at 950 C (1742 F) and cooled in air (slow cooling).The microstructure is similar to Sample AC.Again, a-and j-phases but additionally b 0 are observed.
Coarse j I precipitates and finer j II are formed, and the lamellar j III is mixed with b 0 .The lamellae of the j III are finer than in Sample AC, which can be explained by a higher cooling rate in the smaller laboratory sample compared with an industrial ingot (Figs.3d and 3e).The amount of j IV, is increased because during heat treatment an equalization of the alloying elements can take place caused by diffusion (Fig. 3f).Compared with Sample AC, an increase of the hardness from 160 to 183 HV5 was observed, which is attributed to finer j III , the presence of b 0 , and a higher amount of j IV .

Second Heat Treatment
At first, the NAB samples T2 and T3 were heat treated at 950 C (1742 F) for 1 h and water quenched followed by a second heat treatment.Sample T2 was heated to 700 C (1292 F), kept at this temperature for one day, and water quenched.According to the equilibrium diagram, the b 0 -phase will BO ¨HM ET AL.ON NAB CORROSION decompose to a-and j-phases.After etching, j I and j II are well contrasted, they are colored bluish (Fig. 4a and 4b).The typical morphologies for j III (lamellae) and j IV phase (dots) are not observed (Fig. 4c), but in the large a-grains, a high amount of fine needle-like precipitates are formed and it is impossible to identify the type of j-phase, j III or j IV .The irregular distribution of the "needles" can be explained by different concentrations of the alloying elements in the a-and b-phase during solidification of the NAB melt.Compared with sample T0, a decrease of the hardness from 232 to 197 HV5 was observed.Compared with Sample AC, an increase of hardness from 160 to 197 HV5 was observed.
Sample T3 was heated to 500 C (932 F), kept at this temperature for one day, and water quenched.After etching a-, j I -and j II -phases are well distinguishable (Fig. 4d and 4e).The precipitates of j III -and j IV are very fine and cannot be identified explicitly (Fig. 4f).This important microstructural change results in a ca.18 % increase in hardness for sample T2 in relation to T1.

ELECTROCHEMICAL CORROSION TESTS
Typical results from potentiostatic corrosion testing in SSW are presented for material T0 in simulated sea water (SSW) in Fig. 5a.Qualitatively, potentiostatic data in SSW were similar for all materials and steady state currents were established within ca.24 h.After 70 h, low anodic current densities, i.e., passive or inactive behavior, were established at the test potentials À211 and À156 mV SCE .While the most negative potential (-262 mV SCE ) caused just cathodic currents, the most positive potential (-104 mV SCE ) resulted in 2 orders of magnitude higher anodic currents.
The current densities determined finally, after 70 h of testing, are presented in Fig. 5b for all materials.We may conclude for SSW, there is a sharp transition from passive to active behavior in the potential interval À156 to À104 mV SCE for all samples.Samples T0 and T2, after corrosion testing in SSW at þ 104 mV for 70 h, were selected and their cross sections were prepared for metallographic studies.The results are shown in Fig. 6.
Again, the microstructure of the heat treated T0 sample contains a-and b 0 -phase.The b 0 -phase is metastable and, in addition, contains higher amounts of Al and Fe compared to the a-phase.The images of the cross section show a dramatic change since the b 0 -phase was selectively corroded, up to a thickness of 100 lm, while the a-phase remained intact (Fig. 6a-6c).
Sample T2 was heat treated at 700 C and therefore the b 0 -phase was decomposed into a-and j-phases.By the absence of b 0 , the local differences of electrochemical activities are balanced (Fig. 6d-6f).The images of the cross section show a smaller and more localized corrosion attack.Metallic copper is observed within the corrosion products, indicating dealloying, and the j II -phase remains unattacked (Fig. 6e).
In SFW, a very different reactivity of NAB was observed compared to SSW.A typical result from potentiostatic corrosion testing in SFW is presented for material T0 in Fig. 7a.At the lower test potentials, materials AC and T0 did not fully establish

FIG. 1
FIG. 1 Cu-Al-5 wt.% Ni-5 wt.% Fe phase diagram modified from Ref. [1].The blue line indicates the Al concentration of the alloy investigated.Red lines indicate the various temperatures for the heat treatments.

FIG. 3
FIG. 3 NAB microstructures after solution annealing and cooling.Sample T0 ¼ 1 h 950 C and H 2 O cooling, Sample T1 ¼ 1 h 950 C and air cooling, LOM pictures after etching with Fe 3 Cl and Cu(NH 4 ) 2 Cl 4 solutions, and SEM-BSE image.

FIG. 5
FIG. 5 Potentiostatic measurements in sea water (SSW): (a) current density evolution of sample T0 for different potentials and (b) final current densities for all samples after 70 h.

FIG. 6
FIG. 6 Cross sections of samples T0 and T2 after 70 h potentiostatic testing in SSW at À104 mV SCE .LOM pictures and SEM-BSE images.