Quantitative chemical-structure evaluation using atom probe tomography: Short-range order analysis of Fe–Al
Introduction
Short-range order (SRO) is typically characterised in terms of well-established SRO parameter formalism [1], [2], where the apparent structure is commonly determined by Fourier analysis of the diffuse scattering intensity measured experimentally using bulk, volume-averaged techniques such as X-ray scattering, neutron scattering and Mössbauer spectroscopy. Atom probe tomography (APT), on the other hand, is a direct microscopy technique that provides a unique combination of highly resolved atomistic information, both chemically and spatially in three dimensions (3D), which can be data-mined for quantitative nanostructural information [3], [4]. Developing and applying methods for reliably revealing the SRO in alloys is essential for interpreting phase transformation phenomena and stress–strain response.
In this work, an Fe–18Al (at%) alloy has been investigated after thermal treatment to produce a state of SRO (further details contained elsewhere [5]), where analysis of the experimental APT data has been carried out using the recently developed generalised multicomponent short-range order (GM-SRO) parameters [6], [7]. Details are given in Section 2.3. This new definition is an extension of the pairwise multicomponent SRO (PM-SRO) parameter developed by de Fontaine [2], which itself has roots in the pairwise Warren–Cowley SRO (WC-SRO) parameter for binary systems [1]. GM-SRO analysis can handle not only correlations of pairs of atoms in multicomponent systems (i.e. the PM-SRO), but due to its generalised nature it can also consider multicomponent correlations in multicomponent systems [6], [7] and is, therefore, highly amenable to the highly spatially and chemically resolved tomographic data provided directly by atom probe microscopy [3], containing many millions of atoms.
GM-SRO analysis is carried out on APT data by shell-based counting of the atoms at discrete 3D radial distances, making this atom-by-atom analysis very similar to the calculation of radial distribution functions (RDFs) [8], [9], [10], [11]. While the latter refers to both chemical species-specific ‘partial RDFs’ and pair correlation functions, the former provides a complete set of GM-SRO parameters that statistically describe the atomic architecture. In a ternary A–B–C alloy for example, GM-SRO analysis results in the following set of parameters: , for each shell number m at discrete radii. The parameter α{AB}{BC}, for example, represents the distribution of B- and C-type atoms around A- and B-type atoms. Similarly, the parameter α{A}{ABC} represents the distribution of A-, B- and C-type atoms around A-type atoms. By standard convention, a positive value of a GM-SRO parameter defines co-segregation (clustering) for a particular set of elements in a certain crystallographic shell, whereas a negative value indicates anti-segregation (ordering) of the two sets of elements [6], [7]. Accordingly, this parameter is equal to zero for a random configuration.
A standard definition of SRO with respect to the exact length scale of ordering does not exist, but clearly differs from long-range order (LRO), which refers to periodic repetition of the crystal lattice over larger distances. There also exists a definition for medium-range order (MRO), used within the amorphous materials community, which refers to order that extends over intermediate distances, beyond the second or third atomic shell, e.g. 1–2 nm [12], [13]. SRO, therefore, exists somewhere below these approximate values with the number of coordination (nearest-neighbour) shells depending on the material's lattice parameter.
In addition to RDFs, GM-SRO analysis extends the application of 3D atom probe data from statistical analysis using one-dimensional (1D) Markov chains and the Johnson and Klotz ordering parameter [14], [15], [16], [17], [18] to a more detailed description of SRO that compliments existing capability for LRO and site occupancy investigation [19], [20], [21], [22], [23], [24], [25], [26], [27], [28], [29]. Whilst GM-SRO analysis has previously been carried out on systems containing clustering of solute atoms [7], [30], this work demonstrates its first application towards quantitative assessment of SRO.
Section snippets
Alloy synthesis and preparation
As per [31], a binary Fe–Al alloy with nominal Al content of 18 at% was prepared by vacuum induction melting in a Al2O3 crucible using Fe with 99.99% purity (purified by zone melting to minimise the impurity content) and Al with 99.999% purity followed by solidification in a Cu mould. The actual alloy composition was determined by inductively coupled plasma (ICP) analysis, results shown in Table 1. Specimens for APT were firstly cut from the as-cast alloy material by electrical discharge
APT maps and mass spectra
A tomographic map showing the elemental distributions of Fe and Al is shown in Fig. 1, demonstrating the need for further data mining beyond simple visual inspection. Fig. 2 shows an example of atom probe mass spectra of single- and multiple-hit ions between 5 and 35 Da, with peak assignments labelled directly on the plots. Peaks corresponding to Fe and Al can be seen in the single-hits spectrum, but due to the level of the background signal, peaks for C (including its molecular forms C2 and C3)
Discussion
In the context of the data obtained in the current work, there have been limitations to the analysis using an atom probe approach. The lack of accuracy of the SRO parameters beyond the first coordination shell (Table 3) results from limited detector efficiency and spatial resolution, which have the effect of obfuscating the underlying structural information. This has previously been identified as a limiting factor in the APT analysis of disordered, amorphous metallic glass material using RDFs
Conclusions
Using APT data, quantitative SRO analysis of an Fe–18Al (at%) alloy has been carried out by calculation of the GM-SRO parameters. The key results are summarised as follows:
- •
Chemical species-specific (GM-SRO) parameters have been calculated via shell-based analysis of the experimental three-dimensional atom probe data.
- •
The magnitudes of the GM-SRO values are very small, indicating that definitive signals have been obfuscated by the ‘randomising effect’ of the combination of limited detector
Acknowledgements
RKWM acknowledges the support of the Alexander von Humboldt Foundation through the award of a Humboldt Research Fellowship. Vicente Araullo-Peters is also thanked for help with identification of crystallographic information within the atom probe data.
References (63)
- et al.
Atom probe tomography investigation of heterogeneous short-range ordering in the ‘komplex’ phase state (K-state) of Fe–18Al (at%)
Intermetallics
(2015) - et al.
Ultramicroscopy
(2007) Mater. Sci. Eng. A
(1998)- et al.
Mater. Sci. Eng. A
(1998) - et al.
Intermetallics
(2002) - et al.
Scripta Mater
(1990) - et al.
Acta Metall. Mater.
(1992) - et al.
Ultramicroscopy
(2007) - et al.
Mater. Sci. Eng. A
(1997) - et al.
Intermetallics
(1997)