See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/362495899

On the crystallography of the Mg–Si–Al nitride nuclei and of the

graphite/nitride interface in spheroidal graphite iron

Article in Carbon · August 2022

DOI: 10.1016/j.carbon.2022.07.045

CITATIONS READS

9 225

6 authors, including:

G. Alonso Tomasz Tokarski

IK4-AZTERLAN AGH University of Science and Technology in Kraków
36 PUBLICATIONS 416 CITATIONS 145 PUBLICATIONS 1,362 CITATIONS

SEE PROFILE SEE PROFILE

Doru M. Stefanescu Marcin Górny

The Ohio State University AGH University of Science and Technology in Kraków
424 PUBLICATIONS 7,143 CITATIONS 113 PUBLICATIONS 958 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Doru M. Stefanescu on 19 March 2023.

The user has requested enhancement of the downloaded file.

Carbon 199 (2022) 170–180 https://doi.org/10.1016/j.carbon.2022.07.045
On the crystallography of the Mg-Si-Al nitride nuclei and of the
graphite/nitride interface in Spheroidal Graphite Iron
*
G. Alonso 1, T. Tokarski2, D.M. Stefanescu 3, , M. Górny2, G. Cios2, R. Suarez 1,4
1. AZTERLAN, Basque Research and Technology Alliance (BRTA), Durango, Spain
2. AGH University of Science and Technology, Krakow, Poland
3. The Ohio State University, Columbus, Ohio and The University of Alabama, Tuscaloosa, Alabama, United States of America
4. Veigalan Estudio 2010, Durango (Bizkaia), Spain

ABSTRACT
The role of Mg-Si-Al nitrides in the nucleation of low-sulfur spheroidal graphite Fe-C-Si alloys has been previously
documented. It was demonstrated that when acting as nuclei, in most cases, these nitrides are not singular crystals but are
part of complex inclusions. To further document the complexities of this phenomenon a Fe-C-Si melt with 3.55%C and
3.1%Si was treated with a 2.5Ca1.6Zr1.4Al inoculating alloy and poured into a thermal analysis (TA) cup. Samples
extracted from the TA cup were analyzed through quantitative optical microscopy. Then, Field Emission Gun Scanning
Electron Microscopy was used to identify the nature of the non-metallic inclusions serving as nuclei. It was demonstrated
that the main nucleation sites are Mg-Si-Al nitrides that many times nucleate on sulfides. Also, some Ti carbo-nitrides
were observed growing on the corners of these nitrides. Further, Selected Area Electron Diffraction techniques provided
information on the crystallinity of the nucleus, as well as on the graphite/nucleus interface and on graphite growth in the
immediate vicinity of the nitride. Transmission Kikuchi Diffraction patterns suggest that the nitrides are MgSiN2
polycrystals with orthorhombic lattice. Turbostratic graphite was identified in the proximity of the nucleus. It supports a
better understanding of the growth of graphite on the nucleating phase and brings some clarity to the ongoing debate on
the mechanisms of graphite growth.

Keywords: spheroidal graphite, nucleation, nitrides, turbostratic graphite

1 INTRODUCTION
Melts of multicomponent Fe-C-Si based alloys contain significant amounts of nonmetallic inclusions that can act as
nuclei for graphite, including undissolved graphite, oxides and solid films of oxides [1,2,3,4,5,6,7,8], sulfides
[7,9,10,11,12,13], nitrides [11,14,15,16,17], carbides, and silicates [18]. During liquid metal processing, additional
nucleating inclusions may result from intentional additions of chemicals, as metal-carbon, metal-oxygen and metal-sulfur
reactions generate carbides, oxides and sulfides, respectively. These inclusions can further combine into complex
carbonitrides and oxysulfides. All these inclusions have been shown to act as nuclei for spheroidal graphite (SG). Indeed,
after Jacobs et al. [4] developed the first two-stage nucleation theory for spheroidal graphite on duplex sulfide-oxide
inclusions, having a core made of Ca-Mg or Ca-Mg-Sr sulfides, and an outer shell of complex Mg-Al-Si-Ti oxides,
numerous researchers demonstrated that in commercial SG irons, complex nuclei composed of various inclusions
nucleating on one another were common, leading to sequential (multi-stage) nucleation. More notable is the silicate theory
developed by Skaland et al. [18] that argues that the SG nuclei are hexagonal double layer compounds, with a Mg-Ca
sulfide core surrounded by silicates (MgO SiO2 and 2 MgO SiO2) or oxides that have low potency (large disregistry) on
which epitaxial graphite growth occurs. For a more complete review of the subject the reader is directed to the review
papers [19,20,21,22].
Experimental research also showed that the composition of the base iron exerts a significant influence on the type of
inclusions that become effective as nuclei in SG iron and on the surface density (nodule count) of graphite aggregates.
Igarashi and Okada [15] identified round MgS with a diameter of 1 to 2 µm to which graphite or a Mg-Si-Al nitride was
attached, sometimes also with MgO. Then, Nakae and Igarashi [17] found that at sulfur contents above 0.005 wt.% the SG
nucleus is a spherical MgS, while at S contents lower than 0.0022 wt.% the nucleus is a hexagonal (Mg,Si,Al)N. At higher
sulfur levels, only sulfides are found in the nuclei, while at the lower sulfur level the predominant compound is the
(Mg,Si,Al)N phase (Fig. 1a). Not shown on the graph is that MgO appears present in smaller or larger amounts at all

* Correspondent author; email stefanescu.1@osu.edu; 182 S High Street, Dublin, OH, 43017, USA
sulfur levels, and that at S in excess of 0.08% when the graphite shape becomes lamellar, the MgCa sulfide also includes
Mn in its composition.

Fig. 1. Graphical representation of the influence of the composition of the base metal on the nodule count and the type of compounds
found in the nuclei: a) the effect of sulfur on an iron with 0.053 wt.% Mg for a cooling rate of 22 K/s (adapted after Nakae and
Igarashi [17]); b) the effect of titanium on low-S irons (0.006 - 0.01 wt.% S) with 0.038 to 0.056 wt.% Mg and (data from Alonso et
al. [23]).
Alonso et al. [23] proved that in low-S irons (<0.01 wt.%) the Ti content of the melt also has a significant effect on the
chemistry of the nuclei. The highest number nodule count was found at 0.02 to 0.03 wt.%Ti. As seen in Fig. 1b, the high
nodule count is associated with a high probability of the Mg-Si-Al nitrides. The nuclei also included oxysulfides and
titanium carbonitrides. At lower Ti contents the Ti(CN) inclusions were absent, and the number of complex Mg-Si-Al
nitrides was low. As the Ti content increased, so did the number of Ti carbonitrides, while the number of complex nitrides
decreased at Ti levels in excess of 0.03 wt.%. At 0.036 wt.% Ti no (Mg,Si,Al)N were found as nuclei, as the high content
of titanium precluded nitrogen from combining with other elements. Fig. 1 demonstrates the important role that the
complex nitrides play in increasing the nucleation potential of iron melts. As a high nodule count has been shown to be
beneficial to preventing microporosity occurrence in the last regions to solidify of castings [24, 25], understanding the
nucleation of graphite on this complex nitride is certainly of high interest.
After Mercier et al. [11] reported that Mg-Si-Al nitrides are potential nuclei for SG, some debate ensued over the
crystallography of the nitrides. Solberg and Onsøien [16] concluded that Mg-Si-Al nitrides are based on a hexagonal
crystal structure with a trigonal superstructure and the composition Mg2.5AlSi2.5N6. The parameters of the fundamental
cell deviate only 1-3% from the parameters of the hexagonal AlN. They also stated that there is no obvious
crystallographic similarity between the nitride particle and the graphite, which brought into question the ability of this
compound to act as a nucleus.
It was also argued [26] that effective nucleation requires both low lattice disregistry between the nucleant and the
graphite, and a larger interfacial energy of the nucleant with respect to the liquid than graphite. Unfortunately, quantitative
evaluation of the interracial energy is difficult since it is influenced by many factors including disregistry.
A TEM study of a complex SG nucleus by Laffont et al. [27] concluded that the sequence of formation of the nucleus
was: a facetted center rich in Fe (maybe inherited from inoculation process); then a sulfide-rich Mg around this center,
then a fan-like Ti(CN), and finally a nitride rich in Al, Mg and Si.
A different sequence was documented by Qing et al. [28]. They used qualitative chemical composition analysis to
argue that the complex nucleus of a graphite nodule was composed of a MgO core, with a Mg-Ca sulfide and Mg-Al-Si
nitrides attached on its sides. From ternary diagrams obtained through SEM- Energy-dispersive X-ray spectroscopy (EDS)
analysis they suggested the existence of a large number of Mg-Si-Al nitrides in SG nuclei, at the same level with oxides
and sulfides. Based on the diffraction patterns of the nitride they established orientation relationships between the oxide
and the nitride which support the nucleation of the nitride on the oxide. However, the issue of the elemental composition
of the nitride remained unresolved, as no quantitative information about the phase composition was presented in the paper.
The focus of the experimental work was to elucidate the chemical composition and crystallography of complex nitride
inclusions serving as nuclei for spheroidal graphite. The graphite/nucleus interface, and in particular the nature of the
defects in the graphite in the vicinity of the nucleus also merits special attention as, while the growth mechanisms of
spheroidal graphite have been extensively studied and debated, very limited research was devoted to the nucleus/graphite
interface.
2
2 METHODS AND MATERIAL
A 7000 kW induction furnace was used to produce an iron melt. The charge materials included (all compositions in
wt.%) pig iron (4.30C, 0.65Si, 0.06Mn, 0.05P, 0.026S, 0.019Ti), steel scrap (0.01C, 0.02Si, 0.4Mn, 0.02P, 0.01S,
0.02Cu), SG iron of returns 1 (3.76C, 2.45Si, 0.22Mn, 0.04P, 0.005S, 0.07Cu, 0.022Ti), and SG iron returns 2 (3.76C,
2.45Si, 0.63Mn, 0.04P, 0.006S, 0.36Cu, 0.017Ti). The silicon and carbon levels were corrected through additions of
ferrosilicon (75.09Si, 1.49Al and 0.77Ca) and synthetic graphite to produce a high silicon base iron. The composition of
the melt is given in Table 1.
Table 1. Chemical composition of the base melt (wt.%)
C Si Mn P S Mg Cu Cr Ti Sb
2.55 3.1 0.26 0.037 0.006 <0.005 0.031 0.037 0.02 0.0045

Graphite spheroidization was achieved through the ladle-treatment of the melt with a Mg containing ferrosilicon (50
Si, 5.03 Mg, 2 Ca, 1.33 Lanthanides, all in wt.%). The final iron analysis was as follows: 3.45C, 3.75Si, 0.25Mn, 0.035P,
0.004S, 0.037Mg, all in wt.%.
The Mg-treated iron was poured into thermal analysis (TA) cups to generate cooling curves. A first cup was poured
without inoculation, while the second cup had a Ca-Zr-Al inoculating alloy (wt.% 75Si, 1.4Al, 2.5Ca, 1.6Zr) deposited on
its bottom prior to cup filling. The two TA cups were sectioned, ground and polished for metallographic examination.
Image processing of ten optical microscopy fields was performed with an Image J software to evaluate the graphite nodule
count (number of graphite aggregates/mm2) and the graphite shape (roundness). Several inclusions found in graphite
spheroids from the inoculated cup were then examined through optical microscopy, and with an Ultra PLUS Carl Zeiss
SMT in combination with an X-Max 20 Oxford Instruments Energy-Dispersive X-ray (EDX) detector. Point spectrums
and line scans were applied to analyze the main elements present in the inclusions and to estimate the type of compounds.
Because of the electron-matter interaction volume, the size of the chemical analysis region within SG for measurement
point was estimated to be in the range of 5µm. It allows only for qualitative information about phases composition. Two
graphite spheroids were selected for further processing with a TEM FEI Tecnai TF 20 X-TWIN to examine the details of
graphite/nucleus interface (coherency) and determine crystallographic elements of the nucleus and its chemical analysis.
Transmission Kikuchi Diffraction (TKD) analysis was performed using FEI Versa 3D Field Emission Gun Scanning
Electron Microscopy (FEG-SEM), equipped with Symmetry Electron Backscatter Diffraction (EBSD) camera from
Oxford Instruments. Acquisition was performed at 20kV, beam current of 16nA and step size of 20nm. Diffraction
patterns with resolution of 622x512 were stored for further analysis. For details of the system setup used for TKD
measurements see ref. [29,30]. To extract thin lamellae suitable for TEM investigation the Focused Ion Beam (FIB)
methodology was used. A thin sample was extracted perpendicularly to the longer rectangular edge of the nitride in the
region showing bright contrast on the graphite-nitride interface (Fig. 9b). After lifting with a micro-manipulator, the
lamellae were thinned by Ga+ ions. To avoid excessive damage of the sample surface the last steps was performed using 2
kV ion polishing.

3 EXPERIMENTAL RESULTS

3.1 Thermal analysis, graphite shape and distribution

The efficiency of inoculation in SG iron can be assessed based on its effect on the eutectic undercooling and on the
nodule count and size distribution. As shown in Fig. 2a, melt treatment with the Ca-Zr-Al inoculant decreased the eutectic
undercooling, by increasing the minimum eutectic temperature, TEmin, from 1128℃ for the not-inoculated iron, to 1154℃
for the inoculated one. The recalescence decreased from 18.5℃ to 4℃. Melt treatment increased the nodule count from
155 to 562 per mm2, as well as the roundness from 0.62 to 0.84 (Fig. 3) and significantly displaced the nodule size
distribution to finer diameters, as illustrated in Fig. 2b.

3
 Fig. 2. Effect of inoculation on the solidification of SG iron: a) cooling curves; b) nodule size distribution.

Fig. 3. Effect of inoculation on graphite shape and surface density: a) not inoculated; 155 nodules/mm2; 0.62 roundness ; b) inoculated
with the 2.5Ca1.6Zr1.4Al, 562 nodules/mm 2; 0.84 roundness.

3.2 SEM analysis

A large number of graphite aggregates that exposed nuclei after metallographic preparation were studied by SEM/EDX
to evaluate the chemical composition of the inclusions found to act as nuclei.
The chemical analysis was established by point and line scans analysis (Fig. 4a,b). The elements identified at the two
spots, (1) and (2), are listed on the figure 4a. It can be assumed that both spots detected the presence of a Mg-Si-Al nitride.
On spot (1) a complex La-Ce-Sb oxysulfide is possible. Regarding spot (2), there is not enough information to infer on the
possible compounds that include Zr, Ti and Sb. However, based on previous research [23] it is safe to conclude that (2)
indicates the presence of (TiZr)(CN). Because of the large electron-matter interaction volume detailed analysis of the
white phases surrounding nitride is not possible, however from following line scan analysis presented in fig. 4b spatial
distribution of the elements can be approximated.
The line scans include nitrogen in addition to the previously identified elements. The peaks of the line scans for Mg, Si
and N, labeled (MgSiAl)N, are situated at the same spatial position. Aluminum is also present in the region, but only
exhibits one clear coincidental pick with N, as opposed to Mg and Si that show two peaks each. It is then reasonable to
conclude that the complex nitride includes some Al in addition to Mg and Si, and thus the inclusion is an Mg-Si-Al
nitride. No conclusions can be reached as to the exact composition of the nitride.
A similar analysis for the line scans labeled (MgCeLaSb)S indicate composition peaks at the same spatial position for
Ce, La, Sb and S, which suggests a Ce-La-Sb sulfide. The peaks also suggest the possibility of a Mg-Ce sulfide. From the
line scans alone is not possible to decide whether the SEM image shows a complex Mg-Ce-La-Sb sulfide or two sulfides,
namely (MgCe)S and (Ce,La,Sb)S. Further incertitude is brought by the fact that Nakae and Igarashi [17], in their work on
the effect of the sulfur level on the nature of the nuclei, suggested that MgO was present in smaller or larger amounts at all
sulfur levels. As the spectrum in Fig. 4 also indicates the presence of O, it may be possible that the compounds are not just
sulfides, but rather complex oxysulfides. Antimony is known to produce a naturally occurring sulfide, Sb2S3, several
oxides including Sb2O3, and a nitride, SbN. Thus, other Sb compounds could participate to the complex nucleus.
The line scans marked (TiZr)N on Fig. 4b that exhibit coincidental peaks for Ti, Zr and N allow to conclude that the
polygonal inclusions observed at the two extremities of the complex (Mg,Si,Al)N are Ti-Zr nitrides. However, as
4
thermodynamic analysis supports the formation of (TiZr)(CN) over (TiZr)N it is reasonable to conclude that the inclusion
is a Ti-Zr carbonitride.

Fig. 4. Evaluation of the chemical composition of a graphite nucleus: a) SEM/EDX spot analysis; b) SEM/EDX elemental line scans.
The analysis of 20 SEM fields concluded that 64% of all nuclei observed were nitrides, including Mg-Si-Al and Ti or
Ti-Zr carbonitrides. The other significant components of the nuclei were sulfides. A selection of nuclei showing SG is
given in Fig. 5. In some cases, single or multiple Mg-Si-Al nitrides of rectangular shape appear to be the only component
of the nucleus, as shown on Fig. 5a. It is also seen that the top nitride in the figure exposes the rectangular face of the
hexagonal prism, while the one in the middle exposes the hexagonal face. Random orientation of the graphite platelets
growing on the nitride during early solidification is also observed, as previously reported [37]. This is consistent with
mixture of distorted and ordered graphite regions observed by Hara et al. [31], or with the ‘‘small crystals randomly
twisted’’ observed by Laffont et al. [32]. However, in most cases the Mg-Si-Al nitride is part of a complex nucleus with
Ti carbonitrides (Fig. 5b) or carbonitrides and sulfides (Fig. 5c) nucleated on the faces of the Mg-Si-Al nitride.
The spectrum analysis of the nucleus in Fig. 5c shown in Fig. 6, demonstrates that Ti carbonitrides nucleated on the
top and bottom sides of the (Mg,Si,Al)N phase, as seen in Fig. 6 for spot (2). The Ca-La-Sb sulfides marked by spot (3)
have nucleated on the rectangular lateral sides.

Fig. 5. Typical nuclei found in the experimental iron: a) rectangular Mg-Si-Al nitrides nucleus; b) complex nucleus showing a Mg-Si-
Al nitride with Ti carbo-nitride growing on one face; c) complex nucleus showing a Mg-Si-Al nitride with Ti-Zr carbo-nitride and Ce-
La-Sb sulfide growing on opposite faces.
5
Fig. 6. Magnification of nucleus in Fig. 5c showing the SEM spectrum at different spots; the numbers on the SEM image show the
position at which the analysis spot was focused.

3.3 TEM analysis

Spatial resolution of chemical analysis performed in the SEM is limited to the approx. 5 µm due to electron-matter
interaction volume. Results of the phase analysis is thus imprecise. To further confirm some of the SEM results, two
graphite aggregates with Mg-Al-Si nitride nuclei were studied using the TEM and the TKD techniques. The first graphite
spheroid exhibited another inclusion in its core, while the second one had other inclusions attached to its edges.

3.3.1 Evaluation of inclusions found in the graphite

Fig. 7 presents the SEM and overall view TEM image of the inclusion of the first graphite spheroid. The white line on
Fig. 7a marks the position at which the FIB cross-section was made. Both SEM and TEM views are mutually
perpendicular showing that the inclusion shape is plate-like. The bright field (BF) TEM image presented in Fig. 7b clearly
shows the central rectangular inclusion surrounded by graphite, while on the dark field (DF) image of the same area (Fig.
7c) a small round inclusion in the middle of plate-like inclusion is also visible. Two distinctive features are seen on the
dark field image of the graphite: a) a layered structure of concentric curved plates close to the nucleus (1 on Fig. 7c), and
b) truncated conical columns growing on the curved plates (2 on Fig. 7c). In the right-hand upper quadrant, it is seen that
thick curved plates (3) appear to grow into the columnar cone (2). In the curved region (0) no clear structure is observable.

Fig. 7. Images of first graphite aggregate: a) general SEM view of the SG aggregate; b) bright field image of a graphite quasi-spheroid
with gray rectangular Mg-Si-Al nitride in the center; c) dark field image of graphite in (a) showing the white nitride in the center,
surrounded by elliptical shaped graphite, with outgrowing graphite sectors.

6
 These observations are to a large extent similar to those of Hara et al. [31] whose analysis of TEM bright field images
of SG concluded that the graphite exhibits a three-fold internal structure, with an amorphous-like central region, annular
rings of a layered intermediate region, and an outer region made up of large polygonal crystalline platelets in a mosaic-
like structure.
It is also noted that the shape of the plate-like nucleus affected the growth of the graphite nodule, distorting it to an
ellipsoid rather than spheroid, as seen in Fig. 7a.

Fig. 8. TEM analysis of (Mg,Si,Al)N in graphite: a) (Mg,Si,Al)N with a (MgCa)S inclusion; b) magnification of the region analyzed;
c) SAED pattern of the (MgCa)S inclusion; d) SAED pattern of the (Mg,Si,Al)N inclusion.
Table 2. EDX analysis of spots 1 and 2 on Fig. 8b.
Spot 1 - (Mg,Si,Al)N Spot 2 - (MgCa)S
Element Weight % Atomic % Weight Atomic %
N(K) 34.72 49.76
%
Mg(K) 31.19 25.76 30.86 38.33
Al(K) 3.82 2.84
Si(K) 30.28 21.64 0.74 0.78
S(K) - - 49.90 46.96
Ca(K) - - 18.50 13.93

Magnification of the central region of the inclusions is presented in Fig. 8. The elemental analysis in spot 1 (Table 2)
reveals that the main elements in the plate-like inclusion are, as expected from the previously presented SEM work, Mg,
Si and N, with only a small amount of Al. The data for spot 2 suggest that the inclusion in the middle of the nitride is a
complex Mg-Ca sulfide. The crystallinity of the sulfide was revealed by the diffraction pattern in Fig. 8c. Diffraction
taken from the sulfide shows a complex pattern, consisting of reflections of several grains and faint reflections from the
nitride. The reconstructed reciprocal lattice cross-section matches the FCC lattice with parameter 0.545 nm, consistent
with literature data [4].
Selected Area Electron Diffraction (SAED) processing of the nitride (Fig. 8d) demonstrates that the nitride nucleus has
a crystalline structure. The diffraction was matched to the orthorhombic lattice of (MgSi)N2 phase. Detailed analysis of
the nitride inclusion having low Al concentration is presented for the second graphite aggregate.
A second graphite aggregate containing the complex nitride was further examined by SEM (Fig. 9a) and TEM. A
STEM picture of the thin lamella is presented in Fig. 9c. The sample was then used for TEM and Transmission Kikuchi
Diffraction (TKD) analysis.
Chemical analysis was performed in selected points as shown in Fig. 10. The spectrum of spot 1 further confirms that
the rectangular inclusion is a Mg-Si-Al nitride. The chemical composition of the inclusion expressed in atomic % has very
low concentration of aluminum and is close to the stochiometric composition of the MgSiN2 nitride (Table 3). Inclusions
at the graphite-nitride interface marked with 3 and 4 are very similar and are composed mainly from the rare earth metals
La, Ce and Sb as well as P and S. As Mg-Ce-La-Sb and Ca-La-Sb sulfides were previously identified in nuclei (Fig. 4 and
7
Fig. 6), it is reasonable to assume that spot 3 and 4 detected the presence of La-Ce-Sb sulfides. The La-Ce-Sb sulfides
may also include some iron. Phosphide inclusions growing on the Mg-Si-Al nitride were also found earlier [23].

Fig. 9. Images of the second graphite spheroid: a) back scattered electrons (BSE) SEM image showing a rectangular faceted crystal; b)
FIB cross section through the graphite and the nucleus; c) STEM registered using high-angle annular dark-field (HAADF) detector -
brighter regions means heavier or thicker sample.
Table 3. EDX chemical analysis at spot 1 on Fig. 10
Element Weight % Atomic %
N(K) 31.11 45.71
Mg(K) 33.04 27.98
Al(K) 0.92 0.70
Si(K) 34.91 25.58

Fig. 10. TEM image together with point analysis; Cu and Ga are contaminants coming from the support and gallium ions from cutting
process.
As for the inclusion marked as 2, the Mg, Si, and N are coming from the proximity of the complex nitride. It does not
include S but does show the presence of phosphorus and oxygen. Oxyphosphides of Ce and La are known and have been
previously identified on the corners of Mg-Si-Al nitrides [23]. It is thus concluded that the inclusion is a (LaCe)(OP)
oxyphosphide.
Maps of elemental distribution at the nitride-graphite interface within an area defined by a red rectangle are presented
in Fig. 11. It further supports the presence of a La-Ce-Sb compound. Iron appears as part of an independent inclusion.

3.3.2 Nitride structure analysis

The structure of the nitride was additionally analyzed using the SEM-TKD technique. The lattice of complex Al-Mg-Si
nitride nuclei reported in the literature is based on the hexagonal structure of the AlN phase (wurtzite type structure). The
closest chemical formula that can be derived (Table 2 and Table 3) is matching the stoichiometry of the MgSiN2 phase.
8
 Fig. 11. TEM image together with map of elements around magnesium-silicon nitride

Fig. 12. Analysis of the nitride structure: a) cross-section of the MgSiN2 structure projected along [001] direction with overlaid sketch
of orthorhombic and pseudo-hexagonal unit cells. For clarity, nitrogen atoms are not included, b) TKD pattern from the grain (2) with
overlaid solution; c) TKD orientation maps of the grains; d) stereographic projection of grains orientation showing rotation of ~62°
along the (001) pole.
9
 Also, the compound with this stoichiometry is crystallizing in the oP (orthorhombic primitive) lattice with parameters
a = 0.52697 nm, b = 0.64734 nm, and c = 0.49843 nm. From the geometrical point of view, the orthorhombic unit cell of
MgSiN2 can be obtained by a distortion of the hexagonal lattice of the AlN compound. Fig. 12a shows a cross-section
through the lattice of the MgSiN2 phase projected along the [001] direction, overlaid with the projection of the
orthorhombic unit cell (in black). For clarity of presentation, the nitrogen atoms are not included. It is clear, that position
of the Mg and Si atoms can be also described as pseudo-hexagonal phase (in red), being slightly distorted from the ideal
hexagon position along the b lattice direction. The observed break of the hexagonal symmetry of the AlN lattice is caused
by the distortion introduced by the additional metallic element present in the structure [33].
To confirm the existence of the orthorhombic phase we performed additional diffraction analysis of the nitride using
the Crystallographic Analysis of the Lattice Metric (CALM) software (www.calm-ebsd.com), which allows derivation of
lattice type and parameters of the unknown EBSD or TKD patterns [34]. In the case of the compounds containing light
elements, the estimated accuracy of the lattice parameter determination is better than 3%. Fig. 12b shows the registered
Kikuchi pattern for nitride with overlaid solution. The CALM derived solution is the oS lattice (orthorhombic base
centered) with parameters a = 0.5376 nm, b = 0.3246 nm and c = 0.4980 nm. The CALM reconstructed lattice is different
from those reported in the literature [38,33], by centering and twice smaller lattice parameter b. This discrepancy is caused
by the structure factor of the MgSiN2 phase, where very low intensity (110) and (11̅0) reflections are responsible for
indexation as oS lattice solution. After correction to oP, the lattice parameter is twice higher, b = 0.6492. The TKD
orientation map (Fig. 12c) exhibits three distinctive grains within the nitride. The corresponding grain orientations are
presented on stereographic projections on Fig. 12d. All crystallites share a common direction parallel to the c-direction of
the unit cell, while grains (1) and (3) are rotated by an angle of approx. 62° with respect to the grain (2). Considering
angles of the distorted hexagonal lattice (Fig. 16), identified misorientation will give minimum mismatch between
structures of grains in the growing nitride.

3.3.3 Evaluation of the graphite/nitride interface

As discussed earlier in section 3.3.1, the bright field image of the first graphite spheroid (Fig. 7c) presents structural
characteristics consistent with the multi-stage growth theory of SG, although only the first two stages are clearly outlined.
In the first stage, ellipsoidal graphite grows in direct contact with the nucleus. Several layers are visible, supporting the
curved-circumferential growth mechanism. In the second stage, columnar graphite appears to outgrow on the ellipsoidal
graphite. Sometimes, a column may appear to be the result of a transition from curved-circumferential to columnar
growth, as for sector (3) on Fig. 7c. Nevertheless, this confirms the multi-stage growth of spheroidal graphite discussed in
ref. [35,36]. Still under question is the crystallinity of the graphite at the nucleus/graphite interface defined by Hara et al.
[31] to be “an amorphous-like region”.
To evaluate the crystallinity of the graphite near the nitride, a sequence of Selected Area Electron Diffraction (SAED)
images was generated by moving the aperture across the nitride/graphite interface (Fig. 13). In the first two diffraction
images (1) and (2) very intense signal coming from the nitride obscures diffraction effects from the graphite. After
correction of image contrast and intensity, blurred, ellipse-like effects corresponding to Debye-Scherrer rings of
turbostratic graphite are becoming visible. Once the diffraction aperture is moved towards the graphite region, the
intensity of the rings increases (3), and finally separate diffraction spots from graphite are appearing (4).

Fig. 13. SAED patterns performed sequentially at the nitride/graphite boundary. Circles with numbers are representing position of the
diffraction aperture moved sequentially from the nitride interface (1) to the graphite region (4).
10
4 DISCUSSION
The classic theory of heterogeneous nucleation requires for the inclusion to have low energy of formation, high
stability in the melt and low disregistry. A thermodynamic analysis [37] concluded that the energy of formation of TiN (-
580 kJ/mol) and of AlN (-538 kJ/mol) is higher than that of MgO (-858 kJ/mol), but lower than that of MgS (-294
kJ/mol). As the energy of formation of these nitrides is positioned between that of MgO and MgS, both common nuclei
for SG, it is sufficiently low for nucleation of graphite. It was also argued that some combination of Mg 3N2 with AlN, and
of TiC with TiN, are occurring, as nucleation was documented on (Mg,Al,Si)N and Ti(CN), but was only seldom, if at all,
reported on Mg3N2 or TiC. Thus, the complex Mg -Si-Al nitride appears to be thermodynamically capable to act as a
substrate.
A summary of experimental evaluation of the chemistry of Mg-Si-Al nitrides is presented in Table 4. As early as 1969,
Mercier et al. [11] suggested a composition of 32 Si, 30 Mg, 4 Al and 34 N (in wt. %) and MgSiN2 as a possible formula.
Using TEM diffraction observations, Solberg and Onsoien [16] assigned the Mg2.5AlSi2.5N6 formula to the Mg-Si-Al
nitride, and further identified it as the stoichiometric compound (AlN)6(Mg3N2)5(Si3N4)5 in which the aluminum nitride is
the dominant compound. Although they made 12 individual analyses of the chemical composition, because of the weak X-
ray signals, only the average value was given in the paper, which precludes evaluation of the chemical composition
variance. Using EDX analysis, Laffont et al. [27] found significantly higher numbers for Mg, Si and Al, but much lower
for N.
The EDX elemental analysis of the Mg-Si-Al nitride in this work summarized in Table 4 is closer to that of Solberg
and Onsoien with the exception of Al, which is considerably lower. The closest chemical formula that can be derived is
matching the stoichiometry of the MgSiN2 phase. According to Bruls et al. [38], MgSiN2 can be deduced from AlN by
replacing two Al3+ ions by a combination of Mg2+ and Si4+. This is consistent with the low Al content found in the nitride
in this work, unlike the higher Al in the chemical formula AlMg2.5Si2.5N6 suggested by Solberg and Onsoien.
Table 4. Summary of composition data for the Mg-Si-Al nitride; element percentages are in at%
Element Mercier Solberg Laffont This work
N 34 46.9 14 47.7
Mg 30 23 31.2 26.9
Al 4 6.9 15.3 1.8
Si 32 23.2 36.1 23.6

The finding that the inclusion in the middle of the nitride is a complex Mg-Ca sulfide supports previous claims by
Jacobs et al. [4], Solberg and Onsoien [16], and by Laffont et al. [27] that a sulfide can nucleate Mg-Si-Al nitrides. It can
thus be concluded that the nucleation was initiated by the (MgCa)S inclusion, on which the Mg-Si-Al nitride grew and
then nucleated the graphite aggregate, similar to the sequential nucleation (MgO)/(MgCa)S/(Mg,Si,Al)N previously
described [23]. The graphite spheroid discussed by Qing et al. [28] is different in that the initial nucleus was a MgO
inclusion on which further nucleation of separate (MgCa)S and (Mg,Si,Al)N inclusions occurred.
Solberg and Onsøien [16] concluded that Mg-Si-Al nitrides are based on a hexagonal crystal structure. However,
SAED observation discussed earlier in this paper could not be matched to the hexagonal phase reported by Solberg and
Onsoien. In-depth investigation with EBSD, TKD and CALM (see section 3.3.2) lead to the conclusion that the
(Mg,Si,Al)N is actually an orthorhombic phase made of three distinctive grains (Fig. 12c) that share a common direction
parallel to the c-direction of the unit cell.
Disregistry calculations between graphite and the Mg3N2 nitride [37] found a small disregistry value of 0.59% between
the (010) plane of the cubic Mg3N2 and the (0001) graphite planes. This implies near perfect coherency and hence the
possibility of graphite nucleation on Mg3N2.
Graphite crystallizing in Fe-C-Si alloys is known to belong to the hexagonal crystal system. Rhombohedral structure
was also identified [39,40]. SAED patterns in this work were instrumental in finding that turbostratic graphite grows as a
first layer at the nucleus/graphite interface.
Turbostratic graphite formation has been documented in milled graphite [41], in plasma-enhanced chemical vapor
deposition of carbon nanosheets [42] and in pyrolytic carbon [43]. To the best of our knowledge, the presence of

11
turbostratic graphite in SG iron was first suggested by Pencea et al. [44] who used Wide Angle X-ray Diffraction
(WAXD) experiments to test extracted nodules from Mg treated cast iron. Evidence of turbostratic graphite was claimed.
However, the macroscopic analysis of the WAXD technique did not allow for a more detailed internal study of the SG.
More recently, using X-ray powder diffractometry, Ohbuchi et al. [45] found that from the total mass of graphite in a SG
iron, turbostratic graphite amounted to a surprising 32.2%.
The current finding of turbostratic graphite at the graphite/nucleus interface has implications on the understanding of
both nucleation and growth of graphite. As shown in Fig. 14, nucleation starts with the spreading of defective graphene
layers [46] that may form regions of turbostratic graphite on the nucleus. The subsequent structural ordering of
turbostratic graphite leads to the formation of distorted columns and then wrinkled and straight layers, as already
suggested by Rouzaud and Oberlin [47]. This sequence has been shown to exist in processing by chemical vapor
deposition of carbon nanosheets [42]. It appears that this structural ordering is also valid for SG growing on a nucleus
where the structure becomes more organized with increasing distance from the nucleus (solidification time).

Fig. 14. SEM image of deep-etched low-sulfur superfine interdendritic graphite in an iron sample quenched during solidification
showing the spread of a defective graphene layer on the Ti carbonitride graphite nucleus [46].
Finally, once again it was demonstrated that the shape of the nucleus affects the shape of the graphite aggregate, at
least in the early stages of growth (e.g., Fig. 5a and Fig. 7a)

5 CONCLUSIONS
Earlier thermodynamic analysis and disregistry calculations for Mg3N2 and AlN [37] support the assertion that these
nitrides are possible nucleants for graphite. Thus, a complex nitride such as Mg-Si-Al nitride would be a good nucleant for
graphite. The nucleation ability of this nitride was established by extensive SEM investigations [15,17,23,25], and the
hexagonal structure of the AlN phase (wurtzite type) was confirmed through TEM work [16,28], for Al contents of 4 to
15.3 at% in the Mg-Si-Al nitride. The transmission Kikuchi diffraction analysis presented in this paper shows that low
aluminum (1.8 at%) Mg-Si-Al nitride can have a (MgSi)N2 structure with orthorhombic lattice. To our best knowledge, it
is first time reported that depending on the aluminum content, the nitride nucleus can crystallize in hexagonal or
orthorhombic lattice.
It was also shown that the nitride can have polycrystalline structure with characteristic grain boundaries having small
atomic misfit between adjacent grains.
Attempts to establish the composition of the Mg-Si-Al nitride have been previously published [16,27,37]. Chemical
analysis of the nitride described in this work yielded a composition close to the stoichiometric compound of the MgSiN2
nitride.
It was also demonstrated that the complex Mg-Si-Al nitride nucleated on a Mg-Ca sulfide, following sequential
nucleation (MgCa)S/(Mg,Si,Al)N. This is in line with the previously described sequential nucleation
(MgO)/(MgCa)S/(Mg,Si,Al)N [23].
The nitride nucleus is surrounded by crystalline and/or highly defected graphite, including previously reported graphite
platelets of random orientation (small crystals randomly twisted), mixture of distorted and ordered graphite regions, and
by turbostratic graphite. While the presence of turbostratic graphite in SG was previously revealed through macroscopic
analysis methods, the SAED analysis performed in this research shows that turbostratic graphite can form as a first layer
during graphite growth on the nitride surface.

12
 Graphite growth is affected by the orthorhombic nitride as the graphite shape in the proximity of the nitride is not
spherical but rather ellipsoidal, confirming previous statements to this effect [22,28].

ACKNOWLEDGMENTS
The authors would like to acknowledge Diputación Foral de Bizkaia for supporting this research.

References

[1] R.W. Heine, C.R. Loper, Dross Formation in the Processing of Ductile Iron, Trans. Am. Foundry Soc. 74 (1966) 274–280.
[2] D.R. Askeland, P.K. Trojan, The Approach to Equilibrium and Dross Formation in Nodular Cast Iron, Trans. Am. Foundry Soc.
(1969) 344-352.
[3] P. Poyet, J. Ponchon, Contributions to the study of nuclei for spheroidal graphite in iron for ingot molds, Fonderie 277 (1969) 183.
[4] M.H. Jacobs. T.J. Law, D.A. Melford, M. J. Stowell, Basic Processes Controlling the Nucleation of Graphite Nodules in Chill Cast
Iron, Metals Technology, (Nov 1974) 490-500.
[5] B. Francis, Heterogeneous Nuclei and Graphite Chemistry in Flake and Nodular Cast Irons, Metall. Trans. 10A (1979) 21–31.
[6] M.A. Gadd, G.H.J. Bennett, Physical chemistry of Inoculation in Cast Iron. In The Physical Metallurgy of Cast Iron; Fredriksson,
H., Hillert, M., Eds.; North-Holland: New York, NY, USA, 1984; pp. 99–108.
[7] J. Campbell, A Hypothesis for Cast Iron Microstructures, Metall. Mater. Trans B, 40B (2009) 786-801.
[8] K.L. Ng, H. Sasaki, H. Kimura, T. Yoshikawa, M. Maeda, Heterogeneous Nucleation of Graphite on Rare Earth Compounds
during Solidification of Cast Iron, ISIJ International, 58(1) (2018) 123–131.
[9] R.J. Warrick, Spheroidal Graphite Nuclei in Rare Earth and Magnesium Inoculated. Irons, Trans. Am. Foundry Soc. 74 (1966)
722-733.
[10] K. Stransky, A. Rek, Contribution to the microanalysis of graphite nuclei in spheroidal graphite iron, Giesserei Rundschau no. 9
(1969) 32.
[11] J.C. Mercier, R. Paton, J.C. Margerie, C. Mascré, Inclusions in spheroidal graphite, Fonderie no. 277 (1969) 191-200.
[12] H. Nieswaag, A.J Zuithoff, The occurrence of nodules in cast iron containing small amounts of tellurium, in: 37th Int. Foundry
Congress, Brighton, England (Sep 1970).
[13] M.J. Lalich, J.R. Hitchings, Characterization of Inclusions as Nuclei for Spheroidal Graphite in Ductile Cast Iron, Trans. Am.
Foundry Soc. 84 (1976) 653-664.
[14] C.X. Sun, C.R. Loper, Titanium Carbonitrides in Cast Iron, AFS Trans., 91 (1983) 639-646.
[15] Y. Igarashi and S. Okada, Observation and analysis of the nucleus of spheroidal graphite in magnesium-treated ductile iron, Int. J.
Cast Metals Res. 11 (1998) 83-88.
[16] J.K. Solberg, M.I. Onsoien, Nuclei for heterogeneous formation of graphite spheroids in ductile cast iron, Mater. Sci. Techn. 17
(2001) 1238-1242. dx.doi.org/10.1179/026708301101509313
[17] H. Nakae, Y. Igarashi, Influence of Sulfur on Heterogeneous Nucleus of Spheroidal Graphite, Materials Trans. 43 (2002) 2826-
2831
[18] T. Skaland, T. Grong, T. Grong, A Model for the Graphite Formation in Ductile Cast Iron, Metall. Trans. 24A (10) (1993) 2321–
2345.
[19] T. Skaland, Spheroidal Graphite Nuclei in Rare Earth and Magnesium Inoculated Irons, in: AFS Cast Iron Inoculation
Conference, Schaumburg, Illinois, Sept. 2005, pp 13-30
[20] I. Riposan, T. Skaland, Modification and Inoculation of Cast Iron, in: D.M. Stefanescu (Ed.), ASM Handbook Vol. 1A Cast Iron
Science and Technology, ASM International, Materials Park Ohio, 2017, pp 160-176
[21] I. Riposan, M. Chisamera, S. Stan, The Role of Compounds in Graphite Formation in Cast Iron - A Review, Mater. Sci. Forum,
925 (2018) 3-11
[22] D.M. Stefanescu, G. Alonso, R. Suarez, Recent developments in understanding nucleation and crystallization of spheroidal
graphite in Iron-Carbon-Silicon alloys, Metals, 10(2) (2020) 221; doi:10.3390/met10020221
[23] G. Alonso, D.M. Stefanescu, E. De la Fuente, P. Larrañaga, R. Suarez, The Influence of Trace Elements on the Nature of the
Nuclei of the Graphite in Ductile Iron, Mater. Sci. Forum 925 (2018) 78–85.
[24] G. Alonso, D.M. Stefanescu, E. Aguado, R. Suarez, The Role of Selenium on the Formation of Spheroidal Graphite in Cast Iron,
Metals 11 (2021) 1600. doi.org/10.3390/met11101600
[25] G. Alonso, D.M. Stefanescu, B. Bravo, G. Zarrabeitia, R. Suarez, Nodule Count, End of Solidification Cooling Rate, and
Shrinkage Porosity Correlations in High Silicon Spheroidal Graphite Iron, Minerals, 11 (2021) 155;
doi.org/10.3390/min11020155
[26] T. Mizoguchi, J.H. Perepezko, C.R. Loper, Nucleation During Solidification of Cast Irons, AFS Trans. 105 (1997) 89-94.

13
[27] L. Laffont, A. Pugliara, T. Hungria, J. Lacaze, STEM observation of a multiphase nucleus of spheroidal graphite, J. Mater. Res.
Techn. 9(3) (2020) 4665-4671. doi.org/10.1016/j.jmrt.2020.02.093.
[28] J. Qing, S. Lekakh, M. Xu, D. Field, Formation of Complex Nuclei in Graphite Nodules of Cast Iron, Carbon 171 (2021) 276-
288. doi.org/10.1016/j.carbon.2020.08.022.
[29] T. Tokarski, G. Cios, A. Kula, P. Bała, High quality transmission Kikuchi diffraction analysis of deformed alloys - Case study,
Mater. Charact. 121 (2016) 231-236.
[30] T. Tokarski, G. Nolze, A. Winkelmann, Ł. Rychłowski, P. Bała, G. Cios, Transmission Kikuchi diffraction: The impact of the
signal-to-noise ratio, Ultramicroscopy 230 (2021) 113372.
[31] T. Hara, T. Kitagawa, K. Kuroki, S. Saikawa, K. Terayama, S. Ikeno, K. Matsuda, Morphologies of Some Graphites in Ductile
Cast Irons, Mater. Trans. JIMM, 55(9) (2014) 1500–05.
[32] L. Laffont, R. Jday, J. Lacaze, An Electron Microscopy Study of Graphite Growth in Nodular Cast Irons, Metall. Mater. Trans. A
49A (2018) 1287–94.
[33] D. Jean, L. Yves, L. Jean, Structure de MgSiN2 et MgGeN2, Bulletin de la Société Française de Minéralogie et de
Cristallographie, 93(2) (1970) 153-159.
[34] G. Nolze, T. Tokarski, Ł. Rychłowski, G. Cios, A. Winkelmann, Crystallographic analysis of the lattice metric (CALM) from
single electron backscatter diffraction or transmission Kikuchi diffraction patterns, J. Appl. Cryst. 54 (2021) 1012-1022.
[35] D.M. Stefanescu, G. Alonso, P. Larrañaga, E. De la Fuente, R. Suarez, Reexamination of crystal growth theory of graphite in
iron-carbon alloys, Acta mater. 139 (2017) 109-121.
[36] E. Ghassemali, J.C. Hernando, D.M. Stefanescu, A. Dioszegi, A.E.W. Jarfors, J. Dluhošd, M. Petrenec, Revisiting the graphite
nodule in ductile iron, Scripta Mat. 161 (2019) 66–69
[37] D.M. Stefanescu, A. Crişan, G. Alonso, P. Larrañaga, R. Suarez, Growth of spheroidal graphite on nitride nuclei: disregistry and
crystallinity during early growth, Metall. Mat. Trans. 50A(4) (2019) 1763-1772.
[38] R.J. Bruls, A.A. Kudyba-Jansen, P. Gerharts, H. T. Hintzen, R. Metselaar, Preparation, Characterization and Properties of
MgSiN2 Ceramics, J. Mater. Sci.: Materials in Electronics, 13 (2002) 63-75.
[39] A.N. Roviglione, J.D. Hermida, Rhombohedral Graphite Phase in Nodules from Ductile Cast Iron, Procedia Mater. Sci. 8 (2015)
924-933. doi.org/10.1016/j.mspro.2015.04.153.
[40] J. Qing, V.L. Richards, D.C. Van Aken, Growth stages and hexagonal-rhombohedral structural arrangements in spheroidal
graphite observed in ductile iron, Carbon 116 (2017) 456-469.
[41] J. C. Bokros, Deposition, Structure and Properties of Pyrolytic Carbon. In P. L. Walker (Ed.), Chemistry and Physics of Carbon
Vol. 5, 1969, pp. 1–118. New York, NY: Marcel Dekker.
[42] B.L. French, J.J. Wang, M.Y. Zhu, B.C. Holloway, Evolution of structure and morphology during plasma-enhanced chemical
vapor deposition of carbon nanosheets, Thin Solid Films 494 (2006) 105-109.
[43] R.B. More, A.D. Haubold, J.C. Bokros, Pyrolytic carbon for long-term medical implants, in: B.D. Ratner, A.S. Hoffman, F.J.
Schoen (Eds.), Biomaterials Science, Elsevier Inc., Academic Press, 2013, pp. 209–222.
[44] I. Pencea, D.M. Ştefănescu, R. Ruxanda, F.V. Anghelina, New Aspects Regarding the Structure of Spheroidal Cast Iron Carbon
Inclusions Revealed by WAXD Investigations, Key Eng. Mater. 457 (2011) 120-125.
[45] A. Ohbuchi, K. Otori, K. Hagiwara, Y. Koike, T. Nakamura, Phase Quantification of Regular and Turbo-stratified Graphite in
Cast Iron by X-ray Diffractometry/Rietveld Refinement, ISIJ International 61(8) (2021) 2249–2255.
doi.org/10.2355/isijinternational
[46] D.M. Stefanescu, G. Alonso, P. Larrañaga, R. Suarez, On the stable eutectic solidification of Iron-Carbon-Silicon alloys, Acta
mater. 103 (2016) 103-114
[47] J.N. Rouzaud, A. Oberlin, Structure, microtexture, and optical properties of anthracene and saccharose-based carbons, Carbon,
27(4) (1989) 517-529.

14

View publication stats