Combined thermal, microstructural and microchemical

Volume 85

International Scientific Journal

Issue 2

published monthly by the

June 2017

World Academy of Materials

Pages 49-79

and Manufacturing Engineering

Combined thermal, microstructural and microchemical analysis of solidification of Al25Si3Cu alloy P. Guba a, A. Gesing b, J. Sokolowski a,*, A. Conle a, A. Sobiesiak a, M. Kasprzak c a

Department of Mechanical, Automotive & Materials Engineering, University of Windsor, # 2175 CEI, 401 Sunset Avenue, N9B 3P4, Windsor, Ontario, Canada b Gesing Consultants Inc., 36 Danville Dr., M2P 1J1, Toronto, Ontario, Canada c Department of Power Electronics, Electrical Drives and Robotics, Silesian University of Technology, ul. B. Krzywoustego 2, 44-100 Gliwice, Poland * Corresponding e-mail address: [email protected]

ABSTRACT

Purpose: This paper present thermal and microstructural and microchemical analyses were conducted on the unmodified experimental alloy Al20Si3Cu (B390.1) solidified in the High Temperature Universal Metallurgical Simulator and Analyser (HT UMSA) under atmospheric pressure (0.1 MPa) and a relatively low solidification rate (-1.2 K/s just after end of solidification), for identification of the thermal events during solidification and the phases in the as-cast structure. Design/methodology/approach: The HT UMSA platform, using a low thermal mass stainless steel cup, enabled the acquisition of high resolution thermal analysis data. Design/methodology/approach: A new approach for de-convolution of the first derivative thermal curves allowed detailed thermal and microstructural phase histories to be documented for solidification of Al-Si alloys. Recently developed SEM/EDS methodology allowed to determine composition and distribution of individual phases that are smaller than the X–ray volume. Findings: Simultaneous consideration of thermal microstructural and microchemical information allowed detailed understanding of the series of events that take place during solidification of Al casting alloy with complex chemistry. In our hypereutectic alloy we document growth of Al(1) dendrites and formation of secondary Si(2) and Al(2) phases all at temperatures higher than the binary equilibrium Al-Si eutectic temperature of 850 K. Practical implications: Even at this slow solidification rate detailed understanding of the solidification microstructure requires consideration of non-equilibrium processes during solidification. Originality/value: We propose an original set of hypotheses that consistently explain the observed non-equilibrium solidification behaviour. Proof of these hypotheses is beyond the scope of this work. Keywords: Combined thermal microstructural, microchemical analysis; High Temperature Universal Metallurgical Simulator and Analyser (HT UMSA); B390.1 aluminium alloy; Non-equilibrium solidification

© Copyright by International OCSCO World Press. All rights reserved. 2017

49

P. Guba, A. Gesing, J. Sokolowski, A. Conle, A. Sobiesiak, M. Kasprzak

Reference to this paper should be given in the following way: P. Guba, A. Gesing, J. Sokolowski, A. Conle, A. Sobiesiak, M. Kasprzak, Combined thermal, microstructural and microchemical analysis of solidification of Al25Si3Cu alloy, Archives of Materials Science and Engineering 85/2 (2017) 49-79. MATERIALS

1. Introduction Introduction 1. Hypereutectic Al-Si-X alloys are often used by industry for high performance cast components. High Si contributes to an increase in hardness and wear resistance of the Al-Si alloys, minimizes the thermal expansion coefficient and maintains low density. The Al in the alloy improves the alloy’s thermal conductivity, when compared to cast iron, by factor of 3. These properties allow hypereutectic Al-Si alloys to be used in automotive industry for engine blocks and pistons where tight tolerances and dimensional stability at elevated temperatures are required [1,2]. As-cast hypereutectic Al-Si alloys mechanical and tribological properties are mainly controlled by the primary Si(1) and secondary* Si(2) particle morphology in the alloy microstructure. Slow solidified cast components contain Si(1) particles of size up to 400 µm [3]. HPDC cast components, where higher pressure and cooling/solidification rates are applied, contain primary Si particles of size up to 20 µm [4]. These Si particles are thermally stable, even after 4 h solutions treatment at 480°C and 4 h artificial ageing at 200°C, the size of the Si particles remains almost constant [4]. Large primary Si(1) particles nucleating in the melt poured below liquidus temperature cause accelerated wear of the HPDC injection pistons, and increase the wear of tools in the machining process. Large, hard silicon particles are also crack initiators in cast machine parts during service loading. In consequence the minimization of the primary, Si(1) and secondary, Si(2) particle size becomes an important objective for aluminium casting plants. In the research community, the objective is to develop Al-Si-X alloys with nano-sized Si particles uniformly distributed in aluminium matrix resulting in ultra-high mechanical properties. For development of new structures of Al-Si-X alloys and their processing technology it is first necessary to understand and characterize the thermal events during solidification and identify the phases in the as-cast structure at atmospheric pressure and low solidification rate. Comprehensive thermal and microstructural analysis of the hypereutectic Al-Si alloys were published by Bäckerud et al. [5]. Our research group conducted comprehensive Thermal Analysis (TA) of the solidification process, metallographic and microchemical analysis research on the solidification of the hypereutectic Al>20Si-3Cu (Al>20

50

50

wt.%Si-3 wt.%Cu) alloy, using HT UMSA. The low thermal mass stainless steel foil cup was utilized to obtain the highest quality thermal solidification data unbiased by the thermal inertia of the mass of the crucible [6-9]. The new technique was able to calculate de-convoluted thermal power evolution curves and fraction-complete curves for individual thermal events vs. time and vs. temperature. The elemental composition measurement of phases in the sample microstructure in commonly used SEM-EDX technique is limited to constituents which are significantly larger than the volume in which the characteristic fluorescence x-rays are generated by the electrons scattered from the incident electron beam. In this work, a recently developed variant of an EDX technique is used to overcome the phase constituent’s size limit, and enable the quantitative elemental analysis of smaller phase constituents of sub-micron size. The technique involves the deconvolution of compositions of individual phases from multiple analyses of areas containing phases of interest. This significantly increases the usefulness of EDX for quantitative analysis of new nano-structured materials [10].

2. Experimental 2. Experimental procedures procedures 2.1. Testsample samplechemistry chemistry and design 2.1. Test and design

A hypereutectic Al-Si-Cu ingot with the nominal chemical composition presented in Table 1, was utilized in this research project. Average elemental composition of the tested sample was determined from the mounted and polished section through and normal to the axis of the tested sample. Average elemental composition was measured by two methods: SEM-EDX large area, ~10 mm2 XRF scan, and by hand-held LIBS analyser, SciAps X-300, which ablates material ~300 µm diameter surface spots, and spectrometer analyses the emitted plasma light. X-300 reports Results of these analyses are given in results section, Table 2. This highly-alloyed melt poses very challenging technological requirements as far as the optimization of both as-cast and heat treated structures. In order to study and overcome these challenges, the above-mentioned commercial alloy (hypereutectic Al-Si-Cu) was selected for this research. Special small ingots (donated by the Yamaha

Archives of Materials Science and Engineering

Combined thermal, microstructural and microchemical analysis of solidification of Al25Si3Cu alloy

cast structural gradient with the primary Si, Si(1) concentrated in upper region. In order to limit experimental sample-to-sample chemical variability, all HT UMSA test samples were machined from the centre of the ingot.

Motor Co.) having dimensions of 700 mm x 90 mm x 30 mm (see Figure 1) were designed to limit segregation of the primary Si which nucleates first at the top surface. However, the ingot cross-section analysis revealed a considerable asTable 1. Nominal chemical composition of the experimental alloy Element Si Cu Mg Zn wt.% 20.0 3.0 0.5 0.1

Fe 0.5

Mn 0.1

Ni 0.1

Ti 0.001

Al Balance

SiEQ 21.29

Table 2. Alloy composition Nominal Al Si Mg Fe Cu Mn Cr

min bal. 16 0 0

LIBS max bal. 20 0.5 0.5 3.0 0.1 0.05

0

ave. 68.61% 26.66% 0.38% 0.45% 3.78% 0.08% 0.03%

SEM-EDX st. dev. 12.17% 12.68% 0.27% 0.30% 1.29% 0.04% 0.02%

Ave. st. dev. 66.85% 5.02% 26.59% 3.61% 1.58% 0.62% 0.67% 0.41% 3.55% 0.08% 0.02% 0.02% 0.03% 0.03% selected area averages

Fig. 1. a) Experimental ingot, b) location of the HT UMSA test samples Figure 1 shows the location where the test samples were extracted from the ingot. The dimensions of the HT UMSA test sample and the picture are shown in Figure 2a and b respectively. Sample weight was 17.77 g. The crucibles for melting HT UMSA samples were made from stainless steel foil having a thickness of 0.025 mm. The bottom and top caps were machined from stainless steel having a thickness of 0.25 mm, see Figure 2a and b. The minimal thickness of the crucible walls provides minimal thermal mass which is important for unbiased thermal traces. The crucible foil, caps and the sample were coated with a very thin film of boron nitride spray in order to minimize the reactions between these components, and with the environment.

Volume 85

Issue 2 June 2017

Fig. 2. a) HT UMSA test sample drawing and b) picture

51


2.2. Solidification Solidificationmodelling modelling 2.2.

Solidification of the 71.0 Al, 25.0 Si, 3.0 Cu, 0.5 Fe, 0.5 Mg alloy was modelled using FactSage [12] thermodynamic modelling software Equilib module and light metal alloy database using Scheil solidification option. Equilibrium solidification calculates equilibrium phase assemblies for the starting elemental composition at a series of temperatures spanning the semi-solid region from liquidus to the final solidus. Scheil-Gulliver model assumes that the diffusion in the solid is slow and no reequilibration of solid phases take place during the time frame of solidification. On the other hand, diffusion in the residual liquid is assumed to be fast enough that it can be assumed to be homogeneous throughout and in equilibrium with the currently solidifying metal. In practice Scheil calculation calculates the equilibrium phase assembly for the residual liquid from the previous temperature. In the literature Scheil solidification has been reported to give better match to the observed solidification thermal events and to phase volume fractions than the assumption of full equilibrium throughout the sample during solidification. Scheil solidification model also does not predict phase component shapes and sizes, nor accounts for transport of alloying elements by diffusion or convection, nor for energy barriers to nucleation and growth, nor energy effects associated with local curvatures of the solidifying surfaces. Consideration of these aspects is needed to obtain deeper understanding of the microstructures produced during actual solidification. There is a relatively new development in phase-change computations called phase field modelling that follows the motion of the phase boundaries through the material microstructure at nearly atomic scale. Through combination of thermodynamic and multiphysics approaches in space-time finite difference formulation phase field calculations attempt to account for all the aspects listed above that are missing from the equilibrium and Scheil models. Phase field models are very complex and computationally intensive allowing prediction of structure and phase composition of only very small volumes with practical computation times even on the fastest, largest computer networks. They require many phase property inputs that in many cases have not been measured in detail. Regardless, some of the published results predict alloy solidification microstructures structures that are close to those observed experimentally. In the future combination of experimental solidification thermal, microstructural and microchemical information with phase field model of the process may be able to provide detailed understanding of how the combination of thermodynamic, and mass and heat

52

52

transport phase and elemental component properties lead to the development of a desired alloy microstructure and its evolution during subsequent heat treatment. Experimental data measured in this study could be used to provide model input parameters and provide model validation for a phase field modelling of hyper-eutectic Al-Si-X alloys. This is outside the scope of the present work. We use the Scheil solidification model to give us access to the carefully evaluated thermodynamic database of that in conjunction with the FactSage Equlib free-energy minimization software give us predictions of the phases and the sequence in which the phases are likely to appear, the maximum temperatures at which they can be observed plus their stoichiometry and composition. The liquid and the main metallic phases are thermodynamically modelled as solution phases, and the calculation results predict the evolution of their composition during solidification. On the other hand, the thermodynamic information on the intermetallic phases tends to be incomplete, and in most cases, they are modelled thermodynamically as stoichiometric compounds. In many cases this is in conflict with microchemical analysis results that tend to suggest frequent element substitutions in the intermetallics. We did not perform diffraction experiments and hence we did not identify the crystal structures of the observed phases directly. Rather, we used the intermetallic phase stoichiometry’s and nomenclature from the FactSage light metal database [12]. 2.3. Thermalanalysis-HT analysis-HTUMSA UMSAplatform platform 2.3.Thermal

The High Temperature Metallurgical Simulation and Analysis, HT UMSA, technology platform, shown in Figure 3, can be utilized for identification of characteristic temperatures of the metallurgical reactions during rapid heating, natural cooling, for solution treatment, and artificial aging of the aluminium alloys used in this investigation. The HT UMSA the data acquisition system thermocouples were calibrated of using pure zinc and aluminium samples. Samples produced in HT UMSA tests can then be characterized through metallographic microstructural and SEM-EDX microchemical analysis. Figure 3 shows the system which is made up of: 1) Desktop Computer: used for setup, control, data logging and thermal analysis. 2) Data Acquisition System: national instruments 16-bit system capable of logging temperature/ time measurements on two channels simultaneously with a scan rate of up to 100 Hz per channel. In this work we used one thermocouple logged at 5 Hz. 3) Induction Power Supply: maximum output power of 7.5 kW with the capability to control heating power. 4) Heat Exchange

Archives of Materials Science and Engineering

Combined thermal, microstructural and microchemical analysis of solidification of Al25Si3Cu alloy

System: The UMSA power supply and the electromagnetic coil must be cooled at all times during operation. The heat exchange system provides a steady flow of coolant to the coil. 5) Environmental Chamber contains an integrated induction heating/cooling coil. The chamber provides the

capability to use different heating and cooling modes and rates. Cooling can be performed using either the cooling coil, where gases like argon or nitrogen are blown onto the exterior of the sample, or into the interior of a hollow sample.

Fig. 3. Photograph of the HT UMSA Technology Platform’s main components UMSA Software consists of: 1) UMSA Control/ Monitoring Software is used for experimental control and data logging. It accepts user input in the form of power settings, temperature/time settings, heat treatment paths, etc., and controls the power supply output. If cooling is required, it may also release gas coolants via the integrated cooling coil. 2) Custom Thermal Analysis Software algorithms coded in Octave [13] comprise the postprocessing software used to analyse data logged by the UMSA Control/Monitoring Software. Using the temperature/time data as input the program is capable of calculating information of metallurgical importance, including derivatives, baselines and fraction solid curves as a function of the temperature. The program is well suited for the visualization and comparison of multiple graphs and also has curve smoothing capabilities. In this work the original time-temperature record file was exported for offline thermal analysis as described below.

Volume 85

Issue 2 June 2017

2.4. LOM microstructural analysis 2.4. LOM and SEM/EDS SEM/EDS microstructural analysis Light Optical Microscopy (LOM) analysis was performed using a LEICA DMRE microscope connected to a desktop computer and Leica QWin software. The imaging system was calibrated prior to testing to obtain accurate pixel: micron ratios. A stage micrometre with a 2 mm scale, 0.01 mm sub- divisions, a line width of 2.5 µm and accuracy ± 1.5 µm overall for LOM calibration was used. Phase constituent volume fractions were estimated using image analysis. LOM image analysis was performed using GIMP 2.0 image manipulation software on the digital images. Individual phase constituents were selected by colour, and histogram function provided area count of the selected pixels. Since phase constituents varied drastically in size, magnification was selected to provide trade-off between accurate definition of phase boundary location and accurate capture of the distribution of the

53


phase constituents on the observed metallographic section. For non-uniformly distributed Si(1) crystals that average ~0.5 mm in size we used multiple images with 8 mm horizontal field of view, HFOV, that spanned the entire sample height at the centre, and others spanning the entire sample width at three heights. For more uniformly distributed thin, ~10 µm thick by 100-700 µm diameter Si(2) plates among Al(2) dendrites few images with 0.6 or 1.2 mm HFOV were adequate. For a (Si,Al)(3) structure of 0.5 µm thick Si(3) plates spaced in Al(3) at 1.6 µm 0.12 mm HFOV was necessary to adequately determine the phase boundary location. SEM/EDS analysis was carried out using the FEI Quanta 200 FEG Environmental Scanning Electron microscope equipped with a Field Emission Gun (filament) for highest resolution, Everhart-Thornley Secondary Electron Detector, Solid State Backscatter Detector, Large Field Secondary Electron Detector and EDAX Octane Plus: Energy Dispersive X-Ray Spectroscopy system (EDS) equipped with Silicon Drift Detector, SDD which allows for significantly higher x-ray photon acquisition rates resulting in shorter analysis times and/or better analytical precision. The samples were studied using the SEM in the polished and deep etched surface state in SE and BSE mode and elemental analysis were performed using EDS.

2.5. 2.5. Combined Combinedmodelling, modelling,thermal, thermal, microstructural and analysis microstructural and microchemical microchemical analysis Overall logic of combined modelling, thermal, microstructural and microchemical analysis is as follows: 1. Solidification modelling of the selected alloy composition using Scheil model which assumes mixing and equilibration in the melt and no diffusion or equilibration in the solidified material using FactSage thermodynamic calculation software. This model uses the associated Light Metals thermodynamic database. This database contains some of the best evaluated mixing models for Al alloy melts, solid solutions as well as free energies of formation of all previously observed intermetallic compounds. The Scheil solidification model results summarize the best thermodynamic solidification calculations based on this information and the Scheil assumptions yielding predicted intermetallic compound stoichiometry; solid phase and residual liquid compositions and quantities; as well as the predicted solidification sequence. 2. Sample is solidified in UMSA while collecting temperature-time data in low thermal mass container. 3. Sample polished and observed by LOM to find all visible phases and note their topological inter-relations

54

54

to deduce solidification sequence. Image analysis of multiple images spanning entire height and width of the sample at three elevations at 16X yielded the average volume fraction of primary silicon, Si(1), volume fraction of Si(2) in the binary Al(2)-Si(2) was evaluated in several fields of view at 100X and 200X magnification, the volume fraction of Si(3) in the binary Al(3)-Si(3) structure was evaluated at 1000X magnification. This reflected the decreasing scale of the Si(1), Si(2) and Si(3) structures. 4. Same polished sample is observed by SEM-EDX (SE, BE and EDX) to identify each phase, its composition and to estimate its abundance. This estimate was compared with LOM image analysis results. In many cases the EDX analysis volume was larger than the size of the feature of interest, or the analysis was a selected area average containing two or more phases. The composition and distribution of individual phases in each measurement was calculated through phase composition deconvolution calculation described initially in reference [10]. This calculation was further developed in this work by recognizing that each measurement consists of i independent values of concentration of elements of interest. Since each composition measurement Mik is an average of individual phase compositions Cij weighted by their abundance – distribution Djk. Mik = Cij Djk This set of equations is subject to additional constraints: sum(i, Cij) = sum(j,Djk) = sum(i,Mjk) = 1 and: 0