Study of Thermomechanical Properties of an Al-Zn-Based Composite Reinforced with Dodecaboride Particles

Polymer Engineering Center, Department of Mechanical Engineering, University of Wisconsin-Madison, Madison, WI 53706-1691, USA Department of Materials Science and Engineering, University of Florida, Gainesville, FL 32611, USA Department of Mechanical Engineering, University of Puerto Rico-Mayagüez, Mayagüez, PR 00681-9000, USA Department of Engineering Science and Materials, University of Puerto Rico-Mayagüez, Mayagüez, PR 00681-9000, USA


Introduction
Cast aluminum matrix composites have been widely used in the automotive and aerospace industry due to their weightto-strength ratio [1][2][3][4][5].Because of such application of these types of composites, studying their behavior at high temperatures is mandatory.Researchers demonstrated that the addition of hard boride particles to an aluminum matrix can enhance the composite strength.For instance, AlB 2 particles are an alternative to SiC reinforcement particles in aluminum [6], particularly because SiC/Al composites can have inconsistent wear resistance.
is behavior has been attributed to the weak interface between SiC and the Al matrix [7] with potential deleterious aluminum carbide formation.
By adding hard boride particles to an Al matrix, the wear rate of the composite is lower when compared to unreinforced aluminum [2].In effect, prior studies of Al-Cu alloys reinforced with AlB 2 particles embedded in the matrix demonstrated that diboride contributed to increase effectively the material stiffness [8] with a subsequent ductility loss in the composite.A weak particle/matrix in these composites is a nonissue since it has been reported that the AlB 2 particles can serve as nucleation sites of aluminum upon solidification, which is associated with a low interfacial energy between liquid Al and AlB 2 [6].For structural applications, alloying metals in the aluminum matrix are a necessity to calibrate the final mechanical properties of the material.In particular, Al-Zn-based alloys are amongst the strongest aluminum alloys.For instance, Al alloys bearing up to 30 wt.% Zn can attain discontinuous precipitation at 75 °C, 100 °C, and 150 °C [9].Such precipitation of a second phase improves hardness, Young's modulus, and yield strength of the alloy.Controlled Zn precipitation occurs because of age hardening when precipitation treatment was performed at those low temperatures followed by fast cooling to avoid forming more stable and larger precipitates.Naturally, the resulting microstructure and properties vary and depend on the effectiveness and details of the heat treatment.For this reason, we propose using a Zncontaining Al matrix reinforced with hard aluminum dodecaboride particles, which a prior research proved to strengthen aluminum solid solution matrices [10].
To this purpose, we added AlB 12 particles at different levels into an Al matrix containing 5 wt.%Zn and 10 wt.% Zn as an alternative to other aluminum matrix composites.A thermomechanical analysis (TMA) apparatus, commonly used in the polymer composite field to measure mechanical strength as a function of temperature [11], allowed studying how the composite behaves when subject to high temperatures and heat treatments under compressive loads as done for powder titanium composites [12].Further mechanical and metallographic characterization included Brinell hardness and optical microscopy.

Sample Preparation.
e target chemical composition of the samples was reached by using master alloys bearing Al-95 wt.% Zn, Al-5 wt.% B (containing boron as AlB 12 particles), and aluminum pellets with 99.99% purity.e weighted amounts from each master alloy were calculated according to Table 1 to obtain concentrations of Al-5 wt.% Zn and Al-10 wt.% Zn with 0, 2.08, and 4.17 wt.% B. e corresponding boride volume fraction in the table was computed based on metallographic measurements provided by Melgarejo et al. [2] in a prior research.
All samples were melted at 760 °C in an electric furnace and held at that temperature for 5 minutes for homogenization purposes.At half the holding time, the melt was stirred using a graphite stirrer preheated at 426.7 °C to homogenize the melt further.One should note that the AlB 12 particles remained solid during the entire process as demonstrated by preliminary experimentation and prior research by Adelakin [10].e molten composite was poured into a graphite mold.Each ingot was cut into cubic specimens approximately 5 × 5 × 5 mm using a precision slow saw blade and then polished.e composite specimens were solution treated at 500 °C for 20 minutes in an electrical furnace, followed by quenching into ice water.All specimens were stored in a freezer for 2 to 7 days to avoid natural aging before the TMA experiments.

ermomechanical Analysis.
TMA/SDTA 841 e (Mettler-Toledo AG, Analytical, Switzerland) was used for the thermomechanical studies.For this experiment, the cubic samples were heated to a preset constant temperature, as shown in Table 2, and held at that temperature for 1 hour while a constant compression force of 1 N was applied using a cylindrical quartz probe with a flat cap 3 mm in diameter.
is resulted in a constant compressive stress of 141.47 kPa on the samples for the entire run of the TMA experiment.
e instrument recorded the sample deformation due to the applied load as a function of time during the entire experiment.Negative values are as a result of the compressive test.
After one hour of this heat treatment, the load was removed and the samples were slowly cooled at 2 °C/min to replicate annealing conditions.We computed the deformation of every sample upon loading by subtracting the final thickness (t f ) from the initial thickness (t i ), as shown in the following equation: (1)

Brinell Hardness.
After the TMA experiments, the superficial Rockwell hardness tests were performed on all slowly cooled samples, as we were limited by the small specimen size.Hardness measurements with a 1.5875 mm steel ball indenter were performed onto two parallel, opposed surfaces of each sample.An inverted optical microscope model Nikon EPIPHOT 200 (Nikon Instruments Inc., NY, USA) operated at 5x magnification was allowed for measuring the where P is the force applied on the material surface, D is the indenter diameter, and d is the indentation diameter.

Optical Microscopy.
After fine polishing, a 10% NaOH solution etching for 30 seconds allowed enhancing the microstructure for metallographic analysis.We used the same inverted optical microscope furbished with a highresolution camera to analyze and photograph the microstructure.slow cooling process induced the zinc coarse precipitation because of the decreasing solid solubility of the aluminum matrix.e zinc-rich phase was precipitated on and near the grain boundary in these slowly cooled specimens (nearing equilibrium conditions).As it is well known, this coarse precipitation of an equilibrium secondary phase is not as efficient a hardening mechanism as the fine precipitation of metastable phases (i.e., those pertaining to artificial aging).Moreover, as expected, after the annealing, that is, at room temperature, the composites' hardness increased with higher particle volume fractions.Such hardness improvement results from effective load transfer between the aluminum matrix and the reinforced AlB 12 particles, as previously demonstrated in aluminum-based composites containing B 4 C particles [15,16].AlB 12 particles tend to organize themselves in clusters when surrounded by liquid aluminum, as indicated in previous works, attributed to the lack of sufficient mixing (stirring) when the matrix was molten [17].e same behavior occurred in our experiments due to the short melting time upon sample fabrication.At the same time, for higher concentrations of AlB 12 particles, optical microscopy barely revealed very small pores, which we attributed to the charge material used to incorporate the particles into the melt [10].Upon loading the composite, those pores could induce crack nucleation near or on the particle/matrix interface.Nonetheless, for small porosity levels (as the one observed in the investigated composites), the detrimental effect on the strength of aluminum matrix composites may not be significant when the pores are associated with the hard reinforcements, as indicated by Ahmad et al. [18].

ermomechanical Analysis.
All as-cast samples were annealed for 20 minutes at 500 °C and whereupon quenched in ice water before the TMA experiments.Figure 3 shows the deformation registered by the TMA for samples under compression as a function of time for the Al-5 wt.% Zn (Figure 3(a)) and Al-10 wt.% Zn (Figure 3(b)) composites containing 0, 16.5, and 22.5 vol.%AlB 12 up to one-hour annealing at 125 °C, 160 °C, and 200 °C, respectively.e 5 wt.% Zn samples underwent smaller deformations for higher particle volume fraction at all treating temperatures and higher deformation for samples with no particle volume fraction.Counterintuitively, the 10 wt.% Zn composite bearing the higher particle volume fraction underwent a greater deformation at all temperatures, that is, 125 °C, 160 °C, and 200 °C.Samples with 16.5 vol.%AlB 12 registered the lowest deformation for the 10 wt.% zinc.Naturally, for higher temperatures, the Al-Zn matrix is softened in all samples (higher negative deformation) at the beginning of the TMA experiment.Yet, as time progressed, those deformations decreased mostly for all samples, as shown in Table 3. ese seemingly discrepant results are analyzed later in this section.At this point, we deem noteworthy to mention that these TMA experiments did not activate a creep behavior from any of the tested composites.In effect, the temperatures and load used are not enough for the onset  A multiple linear regression analysis of the experimental data permitted to create a predictive model of the deformation using the Minitab statistical package.
e regression model fitted is shown in (3), and Table 4 displays the fitted parameters.e deformation (in μm) used as a response variable for the model was the maximum deformation difference measured between 500 seconds and 3,600 seconds of the TMA: where a 0 , a 1 , a 2 , a 12 , and a 3 are the fitted regression parameters; B is the boron weight percentage, T is the temperature in °C, and Zn is the zinc weight percent.e selection of the boron weight percent rather than the AlB 12 amount as an independent variable will become apparent in the next paragraphs.
For the fitted model in (3), the very high computed correlation coefficient, that is, R 2 � 0.98, demonstrates the predicting effectiveness of the model.Additionally, the low p values of all fitted parameters indicate that these predictors are all meaningful parts of the model because their eventual changes do effect the response variable.e equation quantifies how the presence of Zn and dodecaboride particles actually increased the compression deformation for every temperature, that is, softening the matrix.As shown in the fitted model, the individual presence of zinc and boron increased the negative (compression) deformation, that is, a 1 > 0 and a 2 > 0. On the other hand, the interaction between both elements (a 12 < 0) reduced the deformation of the composite.e model also indicates, as expected, that temperature is an important factor in holding back the compressive deformation (a 3 < 0). is negative value indicates that temperature is in reality strengthening the aluminum matrix.
e phase diagram in Figure 2 can explain this seemingly contradictory finding.e treatment temperatures indicate that Zn has redissolved into the α Al phase, which led to solid solution strengthening of the matrix.Further analysis of the TMA curves corroborates the fitted model where the slopes of the deformation as a function of time are positive indicating a decrease of deformation with respect to time as shown in Table 3.
Further analysis of the micrographs showed the formation of AlB 2 particles nearby AlB 12 particles for both concentrations of zinc, as shown in Figure 4 for 5 wt.%Zn (Figure 4(a)) and 10 wt.% Zn (Figure 4(b)).
e small amount of diboride particles could not be quantified and is estimated in less than 1 vol.%.On the contrary, when Zn is not present in the matrix, such transformation of the borides particles was not observed.In effect, another study involving Al-based composites revealed that Mg acting as alloying element of the matrix triggered a transformation of the dodecaboride particles.In that case, magnesium induced the decomposition of AlB 12 into Al x Mg 1−x B 2 [20].
erefore, the presence of AlB 2 suggests that zinc favored the AlB 12 to AlB 2 transformation, which somehow might have lowered the strength of the matrix at high temperatures as demonstrated for 10 wt.% Zn where the deformation is greater for samples bearing higher particle volume fraction.As demonstrated by Deppisch et al. [21], by fast cooling the alloy below 1000 °C, nucleation of AlB 12 can be avoided and AlB 2 phase can be nucleated within the alloy.However, the contrary occurs with our samples.By allowing a slow cooling after the TMA analysis, high nucleation of AlB 12 is present with some regions nucleated with AlB 2 .A prior research suggested that the addition of large sized particles may increase the composite ductility [6] at high temperatures, as shown in Figure 3.
e presence of the AlB 12 particles and the addition of zinc into the Al-Zn matrix are inducing a strengthening of the composite matrix, as the deformation decreased for increasing concentration of zinc and boron at high temperatures, as demonstrated in the regression model.

Brinell Hardness.
As mentioned, Brinell hardness (HB) was performed on the samples after they underwent the TMA experiments; this means that the specimens had been slowly cooled (similar to an annealing process) prior to the hardness tests.
e Brinell hardness results in Figure 5 demonstrate how zinc concentration hardened the material after zinc precipitation upon the annealing process, following equilibrium conditions, as shown in the Al-Zn phase diagram in Figure 2. Naturally, HB also increased for higher dodecaboride volume fractions.e average HB values for samples bearing 5 wt.%Zn were 0.158 GPa, 0.184 GPa, and 0.218 GPa for 0, 0.165, and 0.225 vol.%AlB 12 , respectively.Samples containing 10 wt.% Zn obtained HB values of 0.624 GPa, 0.675 GPa, and 0.761 GPa, for 0, 0.165, and 0.225 vol.%AlB 12 , respectively.ese results demonstrate that the material can achieve greater hardness values with the combined addition of Zn as the alloying element of the matrix and the dodecaboride particles as reinforcement at room temperature.Because the dodecaboride particles have higher stiffness than that of the Al matrix, dodecaboride particles produced a constraint resulting in a reduction of plastic deformation of the matrix when HB was performed.As a result, the deformation under compression lowered, which i indicated by a hardness increment.

Discussion
A recent publication proposed a model of grain boundary-(GB-) assisted second phase formation on primary aluminum solid solutions [22], based on a foundational work by Cahn [23].e transformation temperatures are similar to those used in the TMA study of the present research.In a prior research similar transformation temperatures were reported as the authors found that as the temperature fell, so did θ, with θ being the contact angle (dihedral or wetting angle).Figure 6 presents schematically the formation of the Zn-rich β phase (allotriomorphic phase) in the (Al)/(Al) GB with the corresponding dihedral θ angle indicated.In the works by Kogtenkova et al. [24] and Straumal et al. [22,25], for lower temperatures, that is, 125 °C (as in our TMA lower temperature experiments), the dihedral angle becomes zero. is translates into the formation of a thin film of β phase in the GB, which become a virtual Zn sink via lattice diffusion but mostly from GB diffusion.In other words, at lower annealing temperatures, GB diffusion should prevail, leading to dihedral angles nearing zero.
However, in the presence of large irregular particles (as in the present research), the situation becomes more complicated.In particular, aluminum borides are known to have catalytic properties as heterogeneous nucleation sites of primary αAl phase (Al) in aluminum alloys' solidification.
Such catalytic potency is attributed to small disregistry (or mismatch) between low Miller index planes of the boride, and atomic planes of the aluminum nuclei.e precipitated Zn-rich allotriomorphs bearing the said pseudowetting feature could have also formed in the AlB 12 /(Al) interface (Figure 6).
ere are important reasons that could cause    Advances in Materials Science and Engineering such heterogeneous crystallization, one being the strain caused by the differences in coefficient of thermal expansion between the dodecaboride and the aluminum phase.Such strain can be relieved by the formation of new phase on the interface, lowering the overall Gibbs free energy of the material.Unfortunately, the micrographs in Figures 1 and 4 could not be taken at higher magnification and could not prove this potential effect.Nonetheless, the lowest row of micrographs in Figure 1 and the image in Figure 4(b) present a slightly observable feature: the formation of the null wetting angle in the β/(Al) interface.As a consequence, small volume fractions of β phase appear delineating the (Al) grain, as (Zn) formed in the (Al) grain boundaries, as predicted by Cahn.
Once again, one should be careful in analyzing this phenomena independently from the presence of the AlB 12 particle.Such Zn-dodecaboride interactivity is manifest in the thermomechanical properties studied in the present work and has been further revealed by the noticeable effect of the interaction term in the fitted model in (3).
is relationship between microstructure and mechanical properties (or strain) requires a further in-depth study with transmission electron microscopy that is beyond the scope of the present research.
As a whole, the outcome of this experimentation provides a baseline of research for this novel composite.It became evident that it is not enough the addition of dodecaboride particles to formulate a high strength composite if a heat treatable aluminum matrix is used.Not only the amount of dodecaboride particles and the presence of zinc but also the conditions of the treatment and the ensuing cooling rate should be taken into consideration if these particles are to be used as reinforcements for applications at room temperature.

Conclusions
e effect of the addition of AlB 12 particles to gravity cast Al-Zn alloys was studied and related to the composite deformation at high temperature.Deformation of the Al-5 wt.% Zn and Al-10 wt.% Zn matrix composites decreased as a function of time for mostly all temperatures during treatments at 125 °C, 160 °C, and 200 °C with the addition of AlB 12 particles.Brinell hardness after thermomechanometric studies and simulated annealing increased as a function of the AlB 12 particle volume fraction for both Zn concentrations studied.Samples bearing higher Zn levels had higher Brinell hardness due to matrix solid solution strengthening at the temperatures studied.Coarse zinc precipitation in the equilibrium secondary phase occurred for all concentrations and temperatures studied as a result of the slow cooling process after thermomechanical studies were carried out.A multiple linear regression model quantifies the deformation for different temperatures, boron, and zinc levels for these composites.When the influences of zinc and boron (forming aluminum dodecaboride) are considered individually in the predictive model, one can observe that both elements contribute to increase the compression deformation of the composite.However, the model suggests that the interaction between both elements reduce this deformation.

Figure 6 :
Figure 6: Schematic of the formation of β allotriomorphs on the grain boundaries (GB) with the contact angle θ indicated and two possible loci of the AlB 12 particles with incipient heterogeneous β crystallization.Potential diffusion paths via lattice diffusion (J l ) and grain boundary diffusion (J GB ) are also depicted.

TABLE 2 :
TMA treatment temperatures for each concentration.
2Advances in Materials Science and Engineering indentation size with ImageJ public domain image analysis.After measuring each area, the Brinell hardness in GPa was calculated using the well-known following equation: BHN