Content uploaded by Angeliki Lekatou
Author content
All content in this area was uploaded by Angeliki Lekatou on Apr 18, 2015
Content may be subject to copyright.
Partially amorphous stainless steel coatings: Microstructure, annealing behaviour and
statistical optimization of spray parameters
D. Zoisa,b, A. Lekatoua*, M. Vardavouliasb, T. Vaimakisc, A.E. Karantzalisa
aDepartment of Materials Science and Engineering, University of Ioannina, 45111 Ioannina,
Greece
bPyrogenesis SA, Technological Park of Lavrion, 19500, Lavrion, Greece
cDepartment of Chemistry, University of Ioannina, 45111 Ioannina, Greece
*Corresponding author: alekatou@cc.uoi.gr
ABSTRACT
A stainless steel powder of a mixed amorphous and crystalline structure was HVOF
sprayed in an effort to produce coatings with a large glass fraction. In the first part of this work,
the microstructure and annealing behaviour of powder and coatings is studied. The coatings
consisted of a glassy part and a martensitic part, the latter with boride, borosilicide or boro-
carbide dispersions. The annealing behaviour of powder and coatings is characterized by glass
crystallization and martensite tempering. Annealing of the powder leads to complete
microcrystallization of the glassy part, whereas annealing of the coatings eventually leads to
nanocrystallization of the residual glass phase. In the second part, the effects of selected spraying
parameters (oxygen-to-fuel ratio, powder feed rate, spraying distance and spraying stages) on
characteristic coating properties are investigated by means of the Taguchi analysis. The oxygen-
to-fuel ratio mostly affected the coating hardness and porosity. The powder feed rate had a
significant effect on all the coating properties but mostly on the deposition rate and crack
extension force. Spraying in stages significantly increased the deposition rate, whereas it
1
promoted coating amorphicity. A spraying experiment under the optimum conditions determined
by the Taguchi analysis, showed a good fit between the predicted and the attained property
values.
Keywords: HVOF; Stainless steels; Metallic glass; DSC; XRD; Taguchi analysis
Research highlights
•The annealing behavior of partially amorphous steel powders/coatings is clarified.
•Starting powders and coatings show different glass contents.
•Critical (HVOF) spray parameters are optimized by Taguchi Analysis.
•The coating crystalline/amorphous state is correlated with the coating properties.
•The relative fracture toughness of the coatings is evaluated by indentation.
2
1. Introduction
Amorphous alloys have been the subject of significant research efforts due to their
unique physical, mechanical and chemical properties, such as high corrosion and wear
resistance, high hardness, strength, fatigue strength, castability and workability [1]. However,
their increased cost, compared to the polycrystalline conventional materials, has limited their
uses to certain applications [1,2]. A method of reducing the overall cost is the synthesis of
metallic amorphous layers on metal substrates. The manufacture of amorphous coatings is
commonly based on rapid quenching from a melt at cooling rates fast enough to suppress the
nucleation and growth of competing crystalline phases [2]. One method to attain this is to utilize
thermal spray processes. In thermal spraying, powder particles are injected into a hot gas jet,
heated to the molten or semi-molten state and accelerated towards the target-substrate. On
impact with the substrate, droplet spreading occurs, producing lamellar morphologies. Cooling
rates are of the order 107-108 K/s [3]. Thermal spraying is reported as the most economical
method for the production of thick (0.05-2 mm) amorphous coatings [4], which are usually
intended for corrosion and wear resistance applications.
The High Velocity Oxy-Fuel (HVOF) process can be considered as an intermediate
technique between thermal plasmas and cold gas dynamics, as it combines characteristics from
both techniques. Thus, HVOF is characterized by moderate temperatures (lower than those in
plasma spray, higher than those in cold gas dynamics) and high particle velocities (much higher
than those in plasma spray, but still lower than those in cold gas dynamics). The outcome is
coatings with low oxidation and porosity less than 1% [5].
HVOF spraying of amorphous Fe-based coatings has previously been reported. A mixed
amorphous and crystalline structure of Fe-Cr-B-C-Si-Mo-Mn-W coatings has been mentioned by
Chokethawai et al. [6]. Wu et al. [7] deposited Fe-Cr-B-Si coatings consisting of well-flattened
regions, featureless regions and unmelted particles. Ni et al. [8] prepared FeCrMoCBY coatings
3
that exhibited superior microhardness and wear resistance in comparison with electroplated Cr-
and Ni-based amorphous coatings. Wu et al. [9] noted that the formation of borides when
spraying Fe-Cr-Si-B-Mn coatings decreased the boron content in the supercooled liquid and,
hence, its Glass Forming Ability (GFA). Branagan et al. [10] sprayed a Fe-Cr-Mo-B-C-Si-Al
alloy by HVOF and plasma spraying; the produced coatings were primarily amorphous, whilst
they also contained crystalline phases of the 23/6-type (Fe23B6 and Cr23C6). Branagan et al. [11],
also sprayed a Fe-Cr-Mn-Mo-W-B-C-Si powder by HVOF; they claimed that the starting
crystallinity of the finest powder particles does not affect the crystalline state of the resulting
coatings, since the fine particles completely melt and, then, solidify upon impact. However, the
starting crystallinity of the coarsest powder particles significantly affects the crystallinity of the
coatings, since their cores do not melt but are entrained into the coating as solid unmelts. Farmer
et al. [12] reported a very good corrosion resistance for an amorphous HVOF Fe-Cr-Mn-Mo-W-
B-C-Si coating in several environments, including seawater at 90 oC.
A full control of the HVOF process can be a very difficult task since the in-flight particle
characteristics can be affected by multiple parameters, such as the gas-to-fuel ratio, the total gas
flow, gas pressures, the powder feed rate, the spray distance, etc. [13,14]. Kim et al. [5] found
that the amorphous fraction of HVOF Zr-based coatings decreased with increasing the spraying
distance and the fuel flow rate, mostly due to the oxidation of the propelled powder particles.
Fauchais et al. [15] stated that coating reproduction in plasma spraying may be very difficult, as
spraying depends on multiple parameters. Pawlowski [14] noted the similarity of the process
parameters in HVOF and plasma spraying. Examining all the parameters and their influence on
the coating properties is time- and cost-wise ineffective. However, even in the case of managing
to fully control the thermal spray parameters, defects (pores, microcracks, unmelted particles)
might still occur to a small extent [16].
Robust design was firstly introduced as a method that would produce high-quality
products quickly and at a low cost; simultaneously, it would minimize the sensitivity of the
4
production process to variations of uncontrollable factors [17]. These could be achieved by using
proper factorial designs and data analysis. By employing the tabulated basic orthogonal arrays
introduced by Taguchi [18], it is possible to obtain information about the influence of each
parameter through a small number of experiments. A main advantage of using orthogonal arrays
is the use of independent variables; i.e. a change in the value of one parameter does not affect the
effect of the other parameters [17].
Most of the Taguchi experimental design in thermal spraying has been conducted on air
plasma sprayed ceramic coatings. Cizek et al. [19] sprayed hydroxyapatite and evaluated the
effects of parameters, such as particle temperature and velocity, by using an L18 array. Bertrand
et al. [20] sprayed YSZ and evaluated properties, such as porosity, crack length and thermal
conductivity, by using an L16 array.
In the current study, the effects of crucial spraying parameters on characteristic
properties of Fe-based metal glass coatings (MGCs) are examined by Taguchi analysis. The
examined parameters are: (i) oxygen-to-fuel ratio, (ii) powder feed rate, (iii) spraying distance,
(iv) conduction of the spraying process in one or more stages. Their selection was based on the
great influence of these parameters on key physical factors (e.g. melting degree) affecting the
coating properties.
In particular, a close to stoichiometric oxygen-to-fuel ratio has been found to lead to
optimum levels of porosity and hardness of HVOF Ni-5Al coatings [21] and multi-component
white cast iron coatings [22]. It has also been observed to result in maximum values of particle
temperature during HVOF spraying of an alumina powder [23]. An over-stoichiometric ratio has
caused severe in-flight particle oxidation and consequent amorphous phase destabilization
during HVOF spraying of bulk amorphous NiTiZrSiSn feedstock [24]. The oxygen-to-fuel ratio
significantly affects the powder melting degree and, consequently, the amorphous fraction of
HVOF Fe-Cr-Mo-W-B-C-Si coatings [25].
5
The powder feed rate has been reported to seriously affect coating properties, such as
porosity, microhardness, residual stress and modulus of elasticity of plasma sprayed Ni-based
coatings [26]. Lugscheider et al. [27] claimed that a high feed rate results in a low specific heat
per particle and, consequently, a low melting degree and low oxidation during HVOF spraying
of NiCoCrAlY powders.
According to Maranho et al. [22], the spraying distance is a fundamental factor for
HVOF spraying of a multi-component white cast iron. Increasing the spraying distance led to a
decrease in porosity and coating hardness. Turunen et al. [23] noticed that increasing the
spraying distance from 150 mm to 200 mm, led to a large reduction of the particle velocity (~25
%) and a small reduction of the particle temperature (~5 %). As a consequence, increasing the
spraying distance increased the particle melting degree and decreased the thickness per torch
pass.
Spraying in stages is considered to inhibit the excessive rising of the substrate
temperature [14]. Guo et al. [28] observed that a high substrate temperature contributed to the
production of APS (Atmospheric Plasma Sprayed) coatings with a high segmentation crack
density. Furthermore, previous studies [8,29] have claimed that successive torch passes can
induce crystallization from the amorphous material.
Another spraying parameter is the total gas flow that controls the particle velocity [14].
In this investigation, it was decided to employ a fixed total gas flow (that is the maximum one
allowed by the gun) in order to attain a maximum particle deformation through a maximum
particle velocity [13,14,23].
The motivation behind this work was to produce optimized metallic glass coatings
(MGCs) free of expensive rare earth metals by HVOF. The raw material is a stainless steel
nanostructured powder with a great potential to produce coatings of high corrosion and wear
resistance. The studied coatings do not contain expensive rare earth metals, such as Y, Ga and
6
lanthanides, which are usually contained in amorphous steels as glass forming ability enhancers
[30,31]. Additionally, rare earth elements of large atomic diameters are extremely brittle [32].
In this particular effort, critical spraying parameters are optimized by Taguchi analysis, a
method mostly applied to plasma spraying of conventional ceramic powders. The coating
properties selected to be examined, are associated either with the functionality of the coatings
(hardness, porosity, crack propagation resistance) or with the economics of their manufacture
(deposited layer thickness - thickness per pass). In parallel, the evaluation of the relative fracture
toughness of MGCs by indentation is introduced, a technique previously applied to plasma
sprayed ceramic coatings made of nanostructured powders [33] and Fe-based bulk metallic
glasses (BMGs) [34]. Moreover, the responses of the statistical design are elaborated through
findings of a study on the annealing behaviour of the coatings.
2. Experimental
A gas atomized Fe-18.3Cr-7.7Mo-1.6W-1.8Mn-14.9B-3.6C-2.6Si (% at) powder of (-
53+10) μm nominal size, manufactured by the NanoSteel Co. (product code: SHS7574), was
sprayed on 304 stainless steel coupons (40 mm × 40 mm × 5 mm) by a Sulzer Metco DJ2700
Hybrid torch. Propane was used as a fuel. Prior to spraying, the coupons had been degreased and
grit-blasted by corundum particles of (-1 + 0.7) mm grain size.
As fore-mentioned, the examined control factors were: (i) Oxygen-to-fuel ratio (3-
leveled), (ii) powder feed rate (3-leveled), (iii) spraying distance (2-leveled), (iv) conduction of
the spraying process in one stage or three stages (2-leveled). Spraying in one stage included 20
consecutive torch passes. Regarding spraying in three stages, the first stage included 7 torch
passes, the second stage included 7 passes and the third stage included 6 passes; a 2 min cooling
period followed each stage. The control factors are listed in Table I. The experimental
conditions are tabulated in the form of an L9 (34) orthogonal array in Table II. As two of the
7
parameters were 3-leveled and the other two were 2-leveled, the dummy level technique was
used [17]. This technique allows the assignment of a factor of m levels to a column that has n
levels, when n>m. In this case, factor C (spraying distance) and factor D (application or not of
spraying stages) are assigned to three level columns by creating a dummy level (C3 and D3,
where C3≡C1 and D3≡D1). The values assigned to dummy levels are the ones that require to be
extensively studied or are frequently used during the production process.
The factor values, given in Table I, are empirical values commonly employed in the
commercial spraying practice. Especially, each one of the three values of Factor A (oxygen-to-
fuel ratio) corresponds to a specific flame environment (reducing, stoichiometric combustion and
oxidizing). Similarly, the values of Factor B (powder feed rate) are often applied to commercial
spraying. The selected values of factor C (spraying distance) are the ones mostly used in propane
fuel sprayings. Only the procedure regarding level 2 of factor D (spraying in 3 stages) was
specifically conceived for the particular experiments. The cooling intervals between successive
sprayings aimed at reducing the substrate and coating temperatures. This was done by
convection cooling by the compressed air stream coming from the nozzle; cooling was further
reinforced by the substrate rotating motion.
After the realization of the experimental array, the ANOVA (Analysis Of VAriance) tool
was used to interpret the effect of each factor on the coating properties. According to ANOVA,
the effect of each factor on a property is quantified by means of the variance ratio F. The
variance ratio is given by the formula [17]:
e
e
F
F
e
F
SS
DF
DF
SS
V
V
F
==
(1)
where, VF : variation of factor, Ve: variation of error, SSF: sum of squares of factor, SSe:
sum of squares of error, DFF: degrees of freedom of factor, DFe: degrees of freedom of error.
A large value of F means that the effect of the factor is large compared to the error
variance. Also, the larger the value of F, the more important this factor is in influencing the
process response [17].
8
The property (quality) anticipated values under the optimum conditions are estimated
according to the additive model [17]:
opt ij
n n ( n n )
= + −
∑
(2)
Where: nopt is the quality optimum value, n is the mean quality value, i denotes the factor and j
denotes the level.
Practically, a value of F less than 1 means that the factor effect on the property is smaller
than the error; hence, the respective property should not be considered for the additive model. A
value of F greater than 2 means that the factor effect on the property is moderate; hence, the
respective property should be used in the additive model. A value of F greater than 4 means that
the factor effect is quite important.
Here it should be noted that when planning the factorial design, the alternative of
examining the effects of a smaller number of different factors and their interactions was also
considered. However, the intention was to concentrate on the particular four factors for the
reasons reported in the Introduction and attempt to support the experimental responses with the
scientific background extracted from the characterization of the coatings and their annealing
behaviour. Additional involvement of the effects of their interactions would require a much more
extensive matrix, such as an L27 matrix, which is the subject of an ongoing investigation.
The microstructural characterization of the powder feedstock and coatings was conducted
by optical microscopy (Leica DMLM) and SEM/EDX (Philips XL 40 SFEG, JEOL JSM
5600/Oxford Instruments 6587 EDX analyser and Zeiss Supra 35VP/Roentec Quantax (Bruker
AXs) EDX analyser). The coating porosity was estimated by image analysis (Leica AS software)
and Scanning Electron Microscopy (SEM) on cross-sections (×500). It should be emphasized
that the employed image analysis techniques can only identify the coarse pores of thermal spray
coatings and give an overall view of the coating porosity.
9
X-Ray Diffraction (XRD) patterns of powders and surfaces of coatings (as-sprayed and
annealed, whilst unpolished), were obtained by a Bruker D8 Advance diffractometer (Ni-filtered
CuKα1 radiation (λ= 1.5418 Å), standard slit).
Thermal measurements were conducted at a Netzsch STA 449C Jupiter Differential
Scanning Calorimetry (DSC) system (heating rate of 20 oC / min, atmosphere of ultra-high purity
He). The tested coatings had been detached from their substrates and then mortarized. The
sample weight in each DSC run (powders and coatings) was approximately 50 mg.
Besides the standard powder feedstock, its -45 μm fraction was subjected to XRD and
DSC analysis; the -45 μm fraction was received by mechanical sieving of the standard powder
with a No. 325 mesh sieve, in a vibrating apparatus.
Powder and coating annealing was carried out in a HTM Reetz tubular furnace under an
Ar atmosphere.
Microhardness was measured by a Shimadzu tester (14 cross- sectional measurements
per coating). The indentation load was 0.3 kg applied for 10 sec. The mean deposition rate for
each coating was calculated by measuring the coating thickness in five different areas of the
cross-section (×200) and by dividing this with the number of the torch passes. Here, it should be
mentioned that the deposition rate was not determined on the basis of weight measurements,
because the coatings were sprayed on substrates screwed on a vertical rotating cylinder. This
configuration is usually adopted in practice, especially when spraying small components. Hence,
since a fraction of the powder is deposited on the substrate edges, it would be erroneous to
record the weight gain of the substrate.
Crack propagation resistance by indentation is a method having been used to evaluate the
relative fracture toughness of coatings [33,35]. Cross-sections of the coatings were indented by
a Vickers indenter (5 measurements per sample). The indentation load employed was 1 kg
applied for 10 s. The indenter was aligned, so that one of its diagonals was parallel to the
substrate surface. The total length (tip-to-tip) of the major crack (2c) parallel to the substrate
10
surface located at or in the vicinity of the corners of the indentation impression was measured.
The fracture toughness can be extracted from Eq. (3) [36]:
2
3
c
P
K
c
χ
=
(3)
where, Kc: the fracture toughness (MPa*m1/2), P: the indentation load (N), c: the radius of the
radial cracks (m). Coefficient χ can be calculated by Eq. (4):
2
1
=
H
E
ξχ
(4)
where, ξ: a coefficient related to the indenter geometry (for the Vickers pyramid, ξ has been
calculated as equal to 0.016 [36]), E: the Young’s modulus (GPa) and H: the hardness of the
material. H is given by Eq. (5):
2
2
α
P
H
=
(5)
where, α: the impression half-diagonal (m).
The critical crack extension force may be correlated to Kc through Eq. (6) [37]:
E
Kv
G
c
c
2
2
)1(
−
=
(6)
Under plane conditions, combination of Eq. (3-6) leads to Eq. (7):
2
2 2
3
2 (1 )
c
P
G v
c
α
ξ
= −
(7)
where, Gc: the critical crack extension force (J/m2), ν: the Poisson’s ratio (here 0.33, as
determined by previous work for similar bulk metallic glasses [34]).
The use of a value that corresponds to the Poisson’s ratio of Fe-based BMGs (and not
coatings) in Eq. (7) is based on the following considerations: (i) The main reason due to which a
thermally sprayed coating can have a different Poisson’s ratio than the respective bulk material
is the high degree of heterogeneous microstructure (such as pores, interlamellar boundaries,
unmelted particles, oxide stringers, etc.) [38]. In the case of HVOF coatings, the relatively low
11
presence of heterogeneities, as compared to other thermal spraying techniques [5], is expected to
lead to Poisson’s ratio values that are comparatively close to the values of the bulk counterparts.
(ii) Moreover, indentations were conducted on apparently homogeneous areas of coating that
seemed free of discontinuities (at least as observed under the optical microscope). Selection of
such zones has allowed other investigators to involve the Poisson’s ratio of the bulk material in
their calculations when performing indentations on HVOF coatings [39].
The indentation technique has been used to measure the crack extension force of plasma
sprayed ceramic coatings [33,35]. However, as metallic glasses are reported to be very brittle or
have limited ductility [40], it was considered that the method can also be extended to these
materials. Moreover, Fe-based glasses reportedly have poor toughness and are more brittle than
other metallic glasses, such as Zr-based glasses [34].
The porosity of a coating primarily affects its corrosion resistance. The microhardness
and crack propagation resistance are mainly associated with the toughness and wear resistance.
The deposition rate (coating thickness/pass) is mostly related to the cost of coating built-up.
Porosity is in the-smaller-the better category, where an as low as possible value is desired,
whereas the other properties are in the bigger-the better category, where an as high as possible
value is desired [17]. However, the deposition rate should have a maximum limit due to stresses
generated in thick deposition passes [14].
3. Results and Discussion
3.1. Feedstock characterization
3.1.1. Microstructure
Fig.1 (Back Scattered Electron (BSE) mode) shows that the powder particles have a
spherical shape due to the gas atomization processing. Some particles in the shape of a dog bone
12
are observed, owing to incomplete separation of two powder particles during atomization [11].
Most of the particles have size in the range 20-50 μm (Fig. 1). Some finer satellite particles (i.e.
attached to coarser particles) are also discerned. Their formation is caused by back streaming of
the entrained atomization gas during atomization [11]. Most of the coarsest particles present a
dendritic structure. This structure has also been observed in other atomized powders, such as gas
atomized Fe- 44.7 wt% Cr-1.98 wt% Si-5.8 wt% B [9], gas atomized Fe58Cr16Mo8C18 [41], gas
and water atomized Ni-20 wt% Cr [42] and is indicative of crystalline phases.The absence of
texture in the fine particles is indicative of amorphicity [43].
The X-ray diffraction patterns of the powder feedstock, hereinafter reported as “NS”
powder, (as well as the produced coatings) are illustrated in Fig. 2. A hump typical of
amorphicity is noticed; peaks indicating crystallinity are also discerned, suggesting that the
initial powder has a mixed crystalline and amorphous structure. Indeed, previous work has
revealed dispersions rich in W, Mo and Si within the coarsest powder particles [25]. The
assignment of peaks to phases is very complicated, mainly because of the presence of various
types of borides, borocarbides and borosilicides with characteristic peaks at similar angles.
Furthermore, the chemistry of the steel suggests an extensive dissolution of metallic elements in
iron that may cause a peak displacement to slightly different angles than the peak angles of the
simple compositions found in the Powder Diffraction files. For example, Mo, Cr, W and Si (to a
lesser extent) form substitutional solid solutions with αFe in a wide range of compositions [44];
thus, they probably participate in the detected borides and martensite. Indeed, Lin and Hu [45]
have found that the combining tendencies of these elements with boron follow the sequence
Mo>Cr>Fe>Si (as with carbon in steels). Martensite can also dissolve fairly high boron
quantities. The boron solubility in martensite, in Fe-Cr-B cast irons (in the as-cast state), was
measured by Guo and Kelly [46] between 0.185 and 0.515 wt%. These values are much higher
than the accepted solubility value of 0.005 wt% in pure Fe. This significant increase was
associated with alloying element additions, such as Cr and Mo, which, by having atomic
13
diameters larger than that of Fe, expand the iron lattice to sufficiently allow boron atoms to
occupy interstitial sites.
Hence, it should be noted that, owing to the complex composition of the feedstock
powder, the metallic part of the phases encountered in this effort probably consists of solid
solutions of various metals by substitution. For example, phase Fe3B detected by XRD (Fig. 2),
is probably present in the powder as (Fe,M)3B, where M: Cr, Mo and/or W. In fact, (W,Fe)3B
has been reported to form in the system W-Fe-B [47].
The XRD pattern of the -45 μm sieved powder fraction, in Fig. 2, indicates a notably
higher amorphicity than that of the powder feedstock, supporting the fore-mentioned observation
that fine, featureless particles are considered fully amorphous. Thus, it is suggested that
crystalline phases mainly occur in the coarse particles due to a relatively slow solidification after
gas atomization.
3.1.2. Powder annealing
The DSC scans of the standard NS powder feedstock and its -45 μm fraction are shown
in Fig. 3a. Table III gives the peak temperatures and the enthalpies of crystallization extracted
from the respective peak areas. The powder scans show four distinct exothermic peaks. The -45
μm powder fraction exhibits higher exothermic energies on heating compared to the standard NS
powder, suggesting a higher content of amorphous material [8], in compatibility with the
observations of XRD (Fig. 2). The four distinct exothermic peaks correspond to the
crystallization or transformation of different phases. It is also observed that the third peak (Tp3)
and, especially, the fourth peak (Tp4) in the DSC pattern of the -45 μm powder have shifted to
higher temperatures compared to the respective peaks of the standard NS powder. This can be
related to chemistry/phase composition modification. Gas atomized powders show a variety of
microstructures according to their particle diameters corresponding to a degree of undercooling,
with an amorphous structure appearing at high cooling rates [48]. Consequently, phase
14
composition of powder particles may be modified when their size changes during gas
atomization of partially amorphous alloys [48-50]. This differentiation does not only concern the
crystalline part but may also apply to the amorphous part. For example, the amorphous part of
fine particles may present a high density of super-saturated nanoclusters of the base-metal [48].
This phase composition alteration with particle size can be reflected in the DSC patterns of
various particle size ranges by an increase in the peak intensity [48,49] and peak shifting to
higher [48] or lower [49] temperatures with particle size decreasing.
In order to identify the phases corresponding to the DSC peaks, standard NS powder
annealing runs were conducted at the approximate temperatures of completion of the exothermic
peaks, namely 655, 710, 775 and 795 oC, for 2 h.
The XRD patterns of the annealed powders are illustrated in Fig. 4a. More analytically:
(i) Annealing at 655 oC has led to an increase in the intensity of peaks characterizing Fe3B and
Fe15Si3B2, namely 2θ=48.0o (Fe3B), 48.3o (Fe15Si3B2), 43.1o (Fe3B, Fe15Si3B2), 78.4o
(Fe15Si3B2). Some other minor peaks (2θ=44.4o, 2θ=65.1o) may be ascribed to ferrite or Cr-
rich ferrite formed by limited martensite tempering and/or glass crystallization. Here it
should be noted that M3B and M23B6-type structures are quite often reported as the primary
crystalline phases in boron containing Fe-based BMGs [31,45,51-53].Therefore, the lowest
temperature exothermic peak in the DSC spectrum of the standard NS powder (Fig. 3a)
possibly corresponds to the crystallization of Fe3X-type structures from the amorphous
phase, where X: a glass forming element, such as B or (Si,B).
(ii) Annealing at 710 oC has led to an increase in the intensity of peaks characterizing Fe23(B,C)6
– type precipitates (2θ=41.8o, 76.2o, 38.0o) and to a lesser extent Fe23B6 (2θ=37.4o). A small
increase in the presence of αFe {2θ=44.4o, 81.7 o, 64.5 o (PDF: 85-410)} or α(Fe,Cr)
{2θ=44.5ο, 82.0 o, 64.8o (PDF:34-396)} is noted, as drawn from the: (a) Appearance of a
new, hardly discerned peak at 2θ=82.1o. (b) Incorporation of the peak at 2θ=65.5o (assigned
to martensite in the raw powder) into a composite broad peak extending from 64.5 to 65.5o.
15
(c) Significant increase in the intensity of the peak at 44.5o, which, besides being the main
peak of Fe23(B,C)6 and the second highest peak of Cr1.65Fe0.35B0.96, can also be ascribed to
αFe/α(Fe,Cr). Despite the decrease in the respective peak intensity, martensite is still the
main phase of the crystalline part of the powder. In conclusion, the second exothermic peak
in the DSC spectrum of the standard NS powder (at 698 oC) may be assigned to the
crystallization of Fe23B6 and Fe23(B,C)6-type phases (Fig. 3a). The former phase results
from crystallization from the glass. The latter phase may be the direct or indirect outcome
of tempering, as will be further discussed in 3.2.2.
(iii) Annealing at 775 oC has led to a notable decrease in the amorphous fraction of the powder,
since the characteristic hump in the approximate range of 2θ=42o-47o is hardly discerned.
The main peak of the spectrum (2θ=44.6ο) is no longer that of martensite, but the main peak
of αFe or α(Fe,Cr). The composite broad peak extending from 64.5 o to 65.5o, observed in the
spectrum of the specimen annealed at 710 oC, has been replaced by a distinct single peak at
2θ=64.9o. This is better shown in Fig. 4b that presents a magnification of the patterns in the
zone 2θ=60-70o. It is thus inferred that, at 775 oC, martensite transformation by tempering
has been intensified. Peaks clearly assigned to Fe23(B,C)6-type precipitates at lower
temperatures (eg. 710 oC: 2θ=38.1o, 41.8 o, 50.9 o, 51.9 o) have been displaced to slightly
lower angles, which characterize Cr23C6 (ie. 2θ=37.8o, 41.5o, 50.7 o, 51.6 o). At this
temperature, M3(B,Si) and M23B6 type precipitates are no longer observed due to their
metastable state, as further explained in 3.2.2.
At 775 oC, the excess boron that is: (a) produced by the redissolution or transformation of
borides, (b) diffused out of the glass phase and (c) diffused out of martensite (tempered
martensite can dissolve about ten times less boron than as-cast martensite [46]) has reacted
with Cr to form CrB2, a stable compound at high boron contents (> 60% at.) [54].
16
To conclude, the third DSC peak during annealing of the standard NS powder feedstock
(Fig. 3a) may be ascribed to the formation of α(Fe,Cr), Cr23C6 and CrB2 by both intensive
martensite tempering and glass devitrification.
(iv) The XRD pattern of the starting powder annealed at 795 oC does not show any significant
differences in relation to the powder annealed at 710 oC. However, the hardly discerned
hump in the XRD pattern of the powder annealed at 775 oC, has completely disappeared
from the spectrum of the powder annealed at 795 oC, indicating a fully crystalline state. A
well-crystallized Cr-rich ferrite is the primary phase, as the respective intense and sharp
peaks show. The peaks of CrB2 are enhanced, because of the increased diffusivity of Cr and
B. Therefore, the fourth exothermic peak in the DSC spectrum of the standard NS powder
at 786 oC (Fig. 3a) may be attributed to residual glass crystallization, as well as CrB2
precipitation.
3.2. Coating characterization
3.2.1. Microstructure
The measured property values of the coatings are given in Table IV. Comments on the
property values will be reported in section 3.3. Fig. 5 shows a cross-section of the coating
produced from run No. 5 (BSE mode). The coating seems quite dense and of uniform thickness.
Both the minimal presence of discontinuities at the substrate/coating boundaries and the
existence of mechanical interlocking, indicate a good substrate/coating adhesion. The relatively
uniform contrast in the coating suggests a fairly homogeneous structure with occasional presence
of crystalline structure.
Besides the patterns of the standard powder feedstock and its -45 μm fraction, Fig. 2 also
includes the XRD patterns of the as-sprayed surfaces of the coatings produced from runs No. 2,
No. 5 and No. 8. A decrease in the number and intensity of the “crystalline” peaks in relation to
17
the powder feedstock is evident. The coatings exhibit a broad main peak, indicative of
amorphous phases. The hump in the spectrum of the coatings is broader than the hump in the
spectrum of the starting powder. Therefore, it is suggested that, during spraying, a large fraction
of the powder particles was melted and, subsequently, quenched to the glass state. The higher
glass content in the coatings, as compared to the original powder, can be explained by the
quenching of fully molten particles from their highest temperature, through the action of
flattening onto the substrate surface [55]. Kim et al. [56] correlated the high number density of
unmelted particles (defined as the number of spherical particles on a surface of at least 4 mm2)
with a high crystalline phase fraction in Vacuum Plasma Sprayed Ni-Ti-Zr-Si-Sn; they deduced
that crystallization results from the solid state crystallization of partially melted and unmelted
particles. A coarse comparison of the XRD patterns of the three coatings also suggests that
coating No. 5 presents the highest fraction of amorphous phase. The notable presence of
Cr1.65Fe0.35B0.96 in the as-sprayed coatings is justified by the fact that this is a stable phase in the
system Fe-B-X (X: any one or a combination of carbon with one or two of the most common of
the first and second series transition elements) [57]. Similarly, the detection of Fe15Si3B2 instead
of Fe3B in the as-sprayed coatings is justified by the fact that this is a Si-stabilized form of Fe3B
originating from one of the metastable forms of Fe3B [58].
Fig. 6a presents a zone of partially melted particles with a rich dispersion of rectangular
particles (BSE mode). The morphology of the matrix and the XRD analysis indicate a
martensitic structure; EDX analysis showed a high dissolution of hard elements –Cr, Mo, W-
(Fig. 6b). The rectangular precipitates are boro-carbides (Fig. 6c). The large size of precipitates
(often in the micron range) confirms that these were present in the feedstock powder, since
crystallization from an amorphous matrix would lead to notably finer particles of the 20 nm
range [12]. (The presence of Ag in the EDX spectra of Fig. 6 is due to specimen preparation by
Ag sputtering for minimization of surface charging problems on SEM inspection. The same
spectra are the only ones amongst those included in this paper that detect B, despite the
18
appreciable B content in the starting powder, because they are the only ones obtained by an
EDX analyser capable of detecting boron. The expected positions of the elements contained in
the nanosteel powder (as well as of oxygen and silver) are given in Table V.)
Regarding coating No. 8, the appearance of a broad XRD peak (peaking plateau between
35.5o and 36.0o) raises the possibility for the presence of a Fe3O4-type oxide (Fig. 2). Similarly,
multiple oxide stringers are noticed in the cross-sections of all the coatings deposited at the
highest oxygen-to-fuel ratio, as illustrated in Fig. 7a (BSE mode). EDX analysis confirmed the
high oxygen content of the stringers (Fig. 7b).
3.2.2. Coating annealing
Besides the scans of the standard powder feedstock and its -45 μm fraction, Fig. 3a also
includes the DSC scans of coatings from runs No. 2, No. 5 and No. 8. (Coatings No. 2, 5 and 8
were selected on the basis that each one corresponds to a different level of oxygen-to-fuel ratio
and flame environment (sub-stoichiometric/reducing environment, stoichiometric combustion
and over-stoichiometric/oxidizing environment) and, consequently, a different level of melting
degree of the deposited particles [25].) The scans show three exothermic peaks (including a
hardly low temperature discerned peak (Tp1), more clearly shown in Fig. 3b). The three
exothermic peaks correspond to the crystallization or transformation of different phases. In order
to identify these phases, the coatings were subjected to annealing runs at the approximate
temperatures of completion of the exothermic peaks, namely 640 oC, 710 oC, 755 oC (770 oC for
coating No. 2), for 2 h.
Fig. 8 presents the XRD patterns of the surfaces of the annealed coatings. More
analytically:
(i) Annealing at 640 oC has led to the crystallization of Fe3B (only in coating No. 2), Fe3(Si,B),
Fe23B6 and Fe23(B,C)6 – type precipitates from the amorphous part. The primary peak
(2θ=43.4-43.6o) belongs to martensite with the probable participation of Fe23B6. Generally,
19
Fe15Si3B2 is the main phase of the M3B-type borides in the annealed coatings, since, as fore-
mentioned, Fe3(Si,B) is a Si-stabilized form of Fe3B [58].
(ii) At 710 oC, the peak at 2θ=44.6-44.8o has become the main peak along with the peak at
2θ=43.4-43.5o. The appearance of two more peaks at 65.0-65.2o and 82.2o suggest the
presence of α(Fe,Cr) formed by martensite tempering and/or glass crystallization. The
formation of α(Fe,Cr) rather than αFe is attributed to the supersaturation of the disordered
structures of glass and martensite by Cr; upon annealing, the excess Cr diffuses out to form
α(Fe,Cr) and carbides/borocarbides. Peaks corresponding to Fe23(B,C)6 (mainly) and Cr23C6
have increased intensity (i.e. Fe23(B,C)6: 44.4-44.5 o, 37.9-38.0 o, 41.8 o, 50.9 o, 51.8 o, 76.2
o, Cr23C6: 44.2 o, 37.7 o, 41.4-41.5 o, 50.6 o, 51.3 o). Fe23(B,C)6 can be the direct or indirect
outcome of tempering through gradual transformation of Fe23B6 to Fe23(B,C)6 by carbon
diffusion from the martensite lattice to the Fe23B6 lattice. It is, thus, inferred that the second
peak in the DSC patterns of coatings No. 2, No. 5 and No. 8 (Fig. 3a and b) is caused by
the formation of α(Fe,Cr) along with secondary carbides/borocarbides of the 23/6
stoichiometry due to martensite tempering and/or glass crystallization. At this temperature,
the increased instability of glass and martensite has also resulted in the precipitation of
WC1-x, WC and some CrB2 (especially in coating No. 5).
(iii) Annealing at 755 oC (and 770 oC for coating No. 2) has led to a dramatic reduction of the
martensite content by transformation to α(Fe,Cr), as shown in Fig. 8. Cr23C6 particles are
the main precipitates; the increased diffusivity of Cr and C at this high temperature, has
resulted in the stabilization of (Cr,M)23C6. Peaks corresponding to Fe23B6 and Fe3B are not
any longer discerned. Peaks ascribed to (Cr,Fe)2B and Fe3(Si,B) are missing from or hardly
discerned in the coating spectra. In general, the extinction or formation of phases during
annealing at the temperature of the third DSC peak follows the same pattern as with the
powder feedstock, and it is compatible with the well-established facts: (a) M3B and M23B6
are metastable phases [45,51,59]. (b) M23(C,B)6 of variable degree of alloying and
20
morphology (e.g. in some cases virtually pure Cr23C6) is the stable product of tempering
boron alloyed tool steels [60,61]. (c) Cr and Mo rich M23C6 has been detected after
annealing of an amorphous gas atomized Fe58Cr16Mo8C18 powder in the temperature range
678-1052 oC [41].
The observed crystalline phases are also the outcome of glass devitrification, as can been
seen by the large reduction of the characteristic hump. Hence, the presence of stable phases,
such as WC, WC1-x and CrB2 may be attributed to diffusion processes from both the
martensite and the glass phases.
However, although there is not a fourth DSC peak to be ascribed to residual crystallization,
as in the case of the powder, crystallization is not yet complete after annealing for 2 h at
755 oC or 770 oC. A slight hump around the primary XRD peak of α(Fe,Cr) is still
discerned, whereas the broad “crystalline” peaks indicate nanocrystalline phases. (For
example, compare the XRD peaks of α(Fe,Cr) at 64.9-65.0 o and 82.1-82.2 o in the spectra
of Fig. 8 with the clearly sharper respective peaks of α(Fe,Cr) in the spectra of Fig. 4a. The
same for the peaks of Cr23C6 and CrB2.) It is, thus, inferred that the third DSC peak in the
coating patterns (Fig. 3a and b) may be ascribed to intensive martensite tempering and
glass crystallization leading to the formation of α(Fe,Cr), M23(B,C)6, Cr23C6, CrB2, WC.
The residual glass crystallization has rather led to nanocrystalline phases.
3.2.3. Some considerations on the thermal behaviour of the coatings
Nanocrystallization in the annealed coatings can be caused by: (a) crystallization inside
coating defects, such as cracks resulting from relaxation of residual thermal stresses, voids
resulting from poor interlamellar contact or from contact between partially melted or unmelted
particles and previously deposited lamellae [14]. The small size of the defects (for instance, the
gaps due to poor intersplat contact are usually of submicron size) may inhibit crystal growth.
Fig. 9 shows nanocrystals inside a crack, after annealing of coating No. 2 at 770 oC, for 2 h. (b)
21
Upon annealing, heavy elements, such as W and Mo, are rejected from the supersaturated
amorphous matrix and may block growth of nanocrystals [62]. This consideration may be
associated with the presence of clusters of nanoparticles enveloped in a W-rich/Si-poor layer, as
presented in Fig. 10. These nanoparticles are Si-rich, therefore they are the last ones to
crystallize from a relatively stable (Si-stabilized) amorphous matrix; this may be the reason they
are only observed in specimens annealed at the temperature of the third DSC peak. The W-rich
layer could result from diffusion of the excessive W out of the Si-rich glassy matrix, upon
annealing. The outwards diffusion of W is also justified by the fact that Si sharply reduces the
solubility of Mo in α-Fe [63] and probably the solubility of W (Mo and W are fully miscible
[44]). (c) Nanocrystallization by increasing the annealing temperature may be associated with a
much greater increase in the nucleation rate than in the growth rate [62].
Fig. 3a and Table III show that the coatings exhibit lower exothermic energies than the
standard NS powder and its -45 μm fraction, regarding the two low temperature peaks; hence,
the DSC results raise the possibility that the powders contain higher amorphous fractions than
the coatings, contrary to the XRD indications. The above discrepancy has also been noticed by
Branagan et al. [11], on investigating Fe-based amorphous alloys and can be explained on the
basis of their considerations: (i) The mortarized coatings that were subjected to DSC had higher
porosities than the feedstock powders, and, consequently, lower thermal conductivities; the latter
could lead to lower exothermic readings. (ii) There is a high possibility that a thin layer of the
crystalline substrate remained attached to the coatings, after their detachment for the DSC
measurement. Differential Scanning Calorimetry is a bulk analysis technique; as such, DSC
analysis of coatings might have included substrate residues. This would not be possible with X-
ray diffraction, since it is a surface analysis technique.
A comparison of the thermal behaviour of the three coatings, based on Table III, shows
that coating No. 5 exhibits the highest exothermic energies at the low temperature peaks (626
and 699 oC) but the lowest ΔΗ at the temperature of the third peak (736 oC). Fig. 2 indicates that
22
coating No. 5 has a higher amorphous fraction than coatings No. 2 and No. 8. Actually, the XRD
and DSC observations are compatible, since the XRD-based discussion (section 3.2.2) suggested
that only the low temperature DSC peak is clearly associated with crystallization from the
amorphous part. The other two DSC peaks formed at higher temperatures are the resultants of
phase formation both by martensite tempering and residual glass crystallization.
The third peak in the DSC scan of coating No. 2 appears at higher temperature (in
comparison with the other coatings), which approaches the respective peak temperatures in the
powder scans. This could be an indication that the amorphous constituent of the as-sprayed
coating No. 2 is more stable than those of coatings No. 5 and No. 8. Closer observation of the
DSC scan of coating No. 2 in Fig. 3c, raises the possibility that the third peak is formed by the
interpolation of two DSC peaks, peaking at 752 and 764 oC. This may also account for the higher
ΔΗ value in comparison with the respective ΔΗ values of coatings No. 5 and No. 8. Thus, the
resemblance of the DSC scan of coating No. 2 with those of the powders indicates a greater
incidence of unmelted particles (and, consequently, crystalline phases) in this coating, as
compared to coatings No. 5 and No. 8.
To summarize, microstructural observations have suggested that the amorphous part of
the pristine powder is associated with its finest particles. XRD suggests that the coatings have
higher contents of amorphous material than their feedstock powder. Annealing of the powder has
led to full devitrification to a Cr-rich ferrite and stable carbides and borides. Annealing of the
coatings has led to crystallization of the above phases, as well as nanocrystallization of the
residual glass. XRD and DSC suggest that coating No. 5 presents the highest content of
amorphous material, amongst coatings No. 2 and No. 8. DSC arises the possibility that coating
No. 2 contains a higher percentage of crystalline phases than coatings No. 5 and No. 8.
The following section shows that the relative amorphous-crystalline fractions of the
coatings are directly associated with certain spraying parameters and play a primary role in
crucial coating properties.
23
3.3. Factor effects
In Fig. 11, the effect of each factor on the coating properties is illustrated. The values
shown in Fig. 11 represent the mean values of the property measured under the same level of
one control factor, regardless of the levels of the other control factors.
3.3.1. Factor A: Oxygen- to- fuel ratio
3.3.3.1. Hardness
Fig. 11a shows that amongst the investigated factors, the oxygen-to-fuel ratio has a
major effect on the coating microhardness. Previous work [25] showed a direct dependence of
the in-flight particle temperature on the oxygen-to-fuel ratio. Ratios greater than 5 (5: near
stoichiometric combustion) lead to reduced flame temperatures because of oxygen overload;
ratios lower than 5 also lead to decreased flame temperatures because of insufficient oxygen
content [22,25]. The in-flight particle temperature followed the order A1<A3<A2.
The comparatively high flame temperature at the intermediate O2/fuel ratio (≅5), implies
a comparatively high content of melted particles, and, consequently, a high content of
amorphous material after quenching. This is in accordance with the XRD pattern of coating No.
5 in Fig. 2, which suggests a higher amorphous fraction in comparison with coatings No. 2 and
No. 8, as well as with the higher ΔH values of coating No. 5 at the low temperature DSC peaks
(Table III).
The highest microhardness corresponds to the lowest level (A1), which induces the
lowest in-flight particle temperature, as reported in previous work [25]. The relatively low
particle temperature suggests a relatively high content of unmelted (mainly coarse) particles in
the coatings. This claim is enhanced by the observations on the DSC pattern of coating No. 2, in
section 3.2.2. The unmelted / partially melted particles, as illustrated in Fig. 6a are textured.
24
They have a martensitic structure (as suggested by Fig. 2 and the needle-like morphology in Fig.
6a) with a high dissolution of hard metals -Cr, Mo, W- (Fig. 6b). Moreover, they present
dispersions of borides and boro-carbides (as suggested by Fig. 2 and Fig. 6c); their metallic
constituents are rich in refractory metals too (Fig. 6c). Boro-carbides, especially the ones rich in
refractory metals, may tremendously increase the hardness of Fe-based bulk metal glasses.
Khalifa et al. [64] noted that Fe23(B,C)6 is markedly harder than fully amorphous Fe-C-B-Mo-
Cr-W glass. Gan et al. [30] attributed the notably higher hardness of the crystalline alloys, as
compared to their amorphous counterparts ((Fe40Ni40P14B6)100-xGax, x=4-6), to the presence of
hard and refractory intermetallic phases (e.g. (Fe,Ni)23B6, Fe3Ni3B etc.). Hence, the crystalline
(i.e. unmelted) part of the metal glass coatings seems to be harder than the amorphous part, due
to: (a) the martensitic structure of the matrix and (b) the hard particle dispersion; the solid
solution effect of refractory metals significantly enhances the hardening contribution of the
unmelted part. The above claim is further supported by the high hardness of the powder
feedstock: 10 microhardness measurements with a 0.05 kg load in cross-sections of the powder
particles gave an average hardness of 1605 ± 161 HV0.05. Even though the indentation load was
lower than the one used in the microhardness measurements of the coatings (0.3 kg), which
would normally give higher hardness values for the feedstock, the hardness of the feedstock is
almost double the hardness of most of the coatings, probably due to the larger fraction of hard
crystalline phases.
A3 coatings (oxygen-to-fuel ratio: 6.5) have been deposited at lower particle
temperatures than A2 coatings (oxygen-to-fuel ratio: 5). Therefore, they would be expected to be
harder than A2 coatings, owing to their higher fraction of unmelted/crystalline material.
However, they are the least hard of the three types of coatings. This may be attributed to their
higher oxidation content due to the oxygen surplus (1.4 × stoichiometric). Indeed, the XRD
pattern of coating No. 8 is the only one presenting an oxide peak (Fig. 2). Similarly, multiple
oxide stringers are noticed in the cross-section of coating No. 7 (Fig. 7). Three reasons may
25
account for the low hardness of A3 coatings as compared to A2 coatings: (i) Oxides are less hard
than borides or borocarbides. (ii) At the same time, oxides contribute to porosity augmentation
(Fig. 11b) by preventing molten metal from filling asperities. Higher porosities may lead to
lower hardness of coatings, as the carrying load solid area is reduced [65]. (iii) Moreover, brittle
oxides may reduce the overall coating cohesion due to intersplat delamination and oxide
cracking [66]; this might affect the hardness, in case the indentation load penetrates an oxide
stringer /metal interface.
3.3.3.2. Porosity
The lowest porosity has been attained at the oxygen-to-fuel ratio of 5 (A2 level), as
shown in Fig. 11b. A low porosity suggests a high melting degree of particles, as molten
material of high diffusivity has filled the asperities and gaps of the previously deposited layers
[67]. Moreover, in the case of fully melted particles, the in-pressure of the impacting melt
induces mass flow; as a result, well-flattened splats are produced by rapid cooling [56]. Thus, it
is once more confirmed that the near stoichiometric oxygen-to-fuel ratio is responsible for the
highest amorphous-molten material fraction, in compatibility with the XRD, DSC and micro-
hardness measurements.
At the lowest oxygen-to-fuel ratio (3.5: A1 level), many deposited unmelted or partially
melted particles limited or prevented good splashing and gap filling. The contact between
partially melted or unmelted particles and previously deposited lamellae creates voids [14]. One
such void is seen in Fig. 7a (the unmelted particle is marked by a black arrow). On the other
hand, at the highest oxygen-to-fuel ratio (6.5: A3 level), the combination of oxide stringers and
unmelted particles further augmented the porosity of the coatings.
Factor A, has a rather moderate effect on the other two properties (deposition rate and
crack extension force). In each case, the best results were received when the oxygen-to-fuel ratio
was equal to 5. To conclude, the oxygen-to-fuel ratio affects the coating properties mainly
26
because it controls the impacting particle temperature and, consequently, their melting state. As
reported, the melting state of the in-flight particles is the primary parameter that determines the
change in the coating morphology [56].
3.3.2. Factor B: Powder feed rate
The powder feed rate seems to have a significant effect on the coating micro-hardness
and porosity (Fig. 11a and b, respectively). As the material quantity passing through the torch
increases, it is unavoidable that more unmelted particles will be deposited (due to reduced
thermal energy per particle) increasing the coating hardness. Porosity, on the other hand,
decreases with increasing the powder feed rate, due to two factors: (i) The increased incidence of
particles impinging on the pre-deposited material at each torch pass, assists filling any gaps and
asperities. (ii) The increased particle bombardment amplifies the shot peening effect. The shot
peening effect, a characteristic of the HVOF method, produces more compact structures and,
consequently, enhances the coating hardness [7,68,69].
Amongst the measured coating properties, the powder feed rate seems to primarily affect
the deposition rate and the crack extension force. Regarding the deposition rate, it is apparent
that an increased feed rate will result in an increased coating thickness/pass. The crack extension
force is approximately quadruplicated when the feed rate rises from level 1 (24 g/min) to level 3
(38 g/min); this is owing to the more compact structures resulting from the high incidence of
particle impingement. The high number of striking particles per torch pass causes notable
compaction, as particles land on just solidified, hot and, thus, deformable material.
All the produced coatings presented indentation marks rather clear of shear bands (Fig.
12); the latter are mentioned to form in bulk metallic glasses [70].The cracks propagated along
the splat boundaries and parallel to the substrate, indicating anisotropy of the coating properties
(Fig. 12). In the vast majority of the spraying experiments, not any crack formation was noticed
when the indentation load was 0.3 kg.
27
From Fig. 11, it is drawn that level 3 (38 g/min) of factor B (powder feed rate) led to the
most satisfactory results regarding coating micro-hardness, porosity, deposition rate and crack
extension force.
3.3.3. Factor C: Spraying distance
The spraying distance seems to have a moderate effect on the coating porosity and crack
extension force (Gc). A short spraying distance may impede particle re-solidification during their
trajectory in the flame, causing intensive splashing and gap filling and, hence, improved
anchoring of the impinging particles [14]. As a consequence, the coating attains low porosity and
high Gc. However, too short distances should be avoided, since they might cause temperature
rising of the substrate due to the excessive heat of the torch; this could initiate crystallization or
oxidation [71]. The effect of Factor C on the other two coating properties (micro-hardness and
deposition rate) is considered statistically insignificant.
3.3.4. Factor D: Spraying stages
In this investigation, spraying stages were chosen as a control factor for one reason: After
20 consecutive torch passes, the substrate temperature (measured by an IR thermometer) was
higher than 300 oC (oxygen-to- fuel ratio: 5, spraying distance: 220 mm). These temperature
levels, may induce recrystallization in the deposited layers close to the substrate through: (i)
phase nucleation incited by the deposition of successive droplets of molten material, which
release latent heat during solidification [71] and/or (ii) adiabatic recalescence in the deposited
layers due to successive torch passes [29]. The above speculation has been checked by
conducting X-ray Diffraction on the surface side and the substrate side of a detached coating,
sprayed under the fore-mentioned conditions (run No. 6). Indeed, the received patterns,
illustrated in Fig. 13, reveal a crystallization tendency for the substrate side of the coating, owing
to heat accumulation. The main feature of the XRD pattern of the coating top surface is a broad
28
halo peak, whereas the main feature of the pattern of the coating bottom surface is a sharper peak
of higher intensity. Hence, it is confirmed that consecutive spraying (20 consecutive torch
passes) leads to an increased crystallization of the coating part in the vicinity of the substrate.
This is undesirable, not only because the resulting coating has a non-uniform composition, thus
non-uniform properties, but also because the coating presents reduced corrosion protection
abilities. It is established that Fe-based metallic glasses demonstrate better corrosion resistance
than their crystalline counterparts [72,73]; therefore, a large fraction of crystalline phases could
diminish the corrosion protection role of the Fe-based coatings. In fact, a preliminary
investigation on the corrosion properties of partially amorphous Fe-based coatings seems to
confirm this [25].
By breaking the spraying procedure into three stages, it was managed to maintain the
substrate temperature during spraying at levels not exceeding 200 oC.
Fig. 11 shows that factor D mostly affects the deposition rate. Spraying in one stage
resulted in a lower deposition rate but higher Gc, as compared to spraying in three stages. The
successive particle impingements during the twenty consecutive torch passes subjected the hot
previously deposited layers to continuous forging (shot peening effect). Continuous forging may
lead to layer compaction and, thus, low thickness per pass. A large number of compact layers
may benefit the relative fracture toughness of a coating. Factor D has a statistically insignificant
effect on the coating hardness and porosity.
3.3.5. Quantification of factor effects
Table VI presents the property factors calculated by Eq. (1). It is shown that: (a)
amongst the examined coating properties, the oxygen-to-fuel ratio (Factor A) has a major effect
on the coating micro-hardness and porosity and a moderate effect on the deposition rate; (b) the
powder feed rate (Factor B) has a significant effect on all the properties, but mostly on the crack
extension force and deposition rate; (c) the spraying distance (Factor C) has a moderate effect on
29
the crack extension force; (d) the spraying stages (Factor D) significantly affect the deposition
rate. The data of Table VI are generally in accordance with the qualitative observations on Fig.
11.
3.4. Optimum conditions
The final step in this data analysis was to predict the property values under the optimum
spraying conditions (anticipated improvements). To do so, the optimum levels for each control
factor were selected. These are: A2 (oxygen-to-fuel ratio: 5), B3 (powder feed rate: 38 g/min), C1
(spraying distance: 220 mm) and D1 (spraying in stages). Often, the increasing or decreasing
trend of a factor may benefit one property but worsen another; thus, a trade-off between the
desired property values has to be made, based on the application requirements.
To predict the improvements of each property, the additive model (Eq. (2)) was
employed; the anticipated values are given in Table VII. Factors with F values <1 were not
used, because if all the factors were included, the predicted improvement would exceed the
actually realized improvement [17].
Finally, a verification experiment with the optimum parameter settings was conducted.
The attained property values are compared with the predicted ones, in Table VII. The deposited
coating, illustrated in Fig. 14a, exhibited property values fairly close to the ones predicted by the
additive model; thus, the soundness of the latter is confirmed. Moreover, the XRD pattern of the
verification coating indicates a highly amorphous structure (Fig. 14b), enhancing the validity of
this statistical design.
4. Conclusions
30
The present work characterizes HVOF stainless steel coatings of high entropy prepared
under different spraying parameters. It also investigates the effect of spraying parameters
(oxygen-to-fuel ratio, powder feed rate, spraying distance and spraying in stages) on
characteristic coating properties (microhardness, porosity, deposition rate and crack extension
force) by applying Taguchi analysis. The main conclusions drawn are:
1. Both, the starting powder and the produced coatings have a mixed amorphous and
crystalline structure. The crystalline part consists of a martensitic matrix (with a high
dissolution of refractory metals and interstitial boron) and dispersions of boride,
borocarbide and borosilicide particles.
2. The coatings present a higher amorphous fraction than the powder feedstock, suggesting
that during spraying, a large fraction of amorphous and crystalline powder particles was
melted and, then, quenched to the glass state.
3. Isothermal annealing of the standard (unsieved) powder feedstock at the temperatures of
the four DSC exothermic peaks demonstrated the formation sequence: Fe3X (X: B or
(Si,B)) by glass crystallization → Fe23Y6 (Y:B or (B,C)) by glass crystallization →
α(Fe,Cr), Cr23C6, CrB2 by glass crystallization and martensite tempering →CrB2
precipitation and residual glass crystallization
4. Isothermal annealing of the coatings at the temperatures of the three DSC exothermic
peaks demonstrated the formation sequence: Fe3X (X: B or (Si,B)) and Fe23Y6 (Y:B or
(B,C)) by glass crystallization → α(Fe,Cr) and M23Y6 (M: (Fe,Cr), Y:(B,C) or C) by glass
crystallization and martensite tempering → α(Fe,Cr), M23(B,C)6, Cr23C6, CrB2 and WC by
martensite tempering and glass crystallization; the residual glass crystallization has led to
a nanocrystalline structure.
5. Nanocrystallization in the annealed coatings has been observed to be associated with: (a)
crystallization inside coating defects, such as stress relaxation cracks, (b) blocking of Si-
31
rich nanocrystal growth by heavy metals rejected out of a supersaturated amorphous
matrix.
6. Amongst the investigated coating properties (hardness, porosity, deposition rate, crack
extension force), the oxygen-to-fuel ratio mostly affects the coating hardness and
porosity. The highest coating hardness corresponds to the lowest oxygen-to-fuel ratio due
to the highest crystalline (i.e. unmelted) fraction. The high hardness of the crystalline
phases is attributed to the presence of borides, borosilicides and borocarbides, as well as
the dissolution of refractory metals in martensite. The lowest hardness corresponds to the
highest oxygen-to-fuel ratio, owing to the highest fraction of oxides. A near
stoichiometric oxygen-to-fuel ratio promotes a high melting degree of particles, resulting
in coatings of high amorphous fractions. As a consequence, the lowest coating porosity is
attained at the near stoichiometric oxygen-to-fuel ratio.
7. The powder feed rate has a significant effect on all the investigated coating properties
and, especially, on the deposition rate and crack extension force. Increasing the feed rate
will increase the coating thickness per pass and the relative fracture toughness; the latter
is attributed to the more compact structures arising from the high incidence of particle
impingement on hot pre-deposited material.
8. A moderately short spraying distance has a positive (modest though) effect on the
coating porosity and relative fracture toughness, by limiting re-solidification.
9. Spraying in stages inhibits the excessive rising of the substrate temperature and,
consequently, benefits the coating amorphicity. Amongst the investigated properties,
spraying in stages significantly affects only the deposition rate by increasing it.
10. The property values predicted by the “additive model” under the optimum control
factors, as determined by the L9 array data, are in close proximity to the respective
experimental values.
32
Acknowledgements
D. Zois is grateful for the financial support by the Hellenic State Scholarship Foundation.
33
References
[1] A.I. Salimon, M.F. Ashby, Y. Brechet, A.L. Greer, Mater. Sci. Eng. A 375-377 (2004)
385-388.
[2] W.H. Wang, C. Dong, C.H. Shek, Mater. Sci. Eng. R 44 (2004) 45-89.
[3] A.H. Dent, A.J. Horlock, D.G. McCartney, S.J. Harris, Mater. Sci. Eng. A 283 (2000) 242-
250.
[4] C. Moreau, P. Cielo, M. Lamontagne, S. Dallaire, J.C. Krapez, Surf. Coat. Technol. 46
(1991) 173-187.
[5] H.J. Kim, K.L. Kim, B.G. Seong, C.G. Park, J. Mater. Sci. 36 (2001) 49-54.
[6] K. Chokethawai, D.G. McCartney, P.H. Shipway, J. Alloys Compd. 480 (2009) 351-359.
[7] Y. Wu, P. Lin, Z. Wang, G. Li, J. Alloys Compd. 481 (2009)719–724.
[8] H.S. Ni, X.H. Liu, X.C. Chang, W.L. Hou, W. Liu, J.Q. Wang, J. Alloys Compd. 467
(2009) 163–167.
[9] Y. Wu, P. Lin, C. Chu, Z. Wang, M. Cao, J. Hu, Mater. Lett. 61 (2007) 1867–1872.
[10] D.J. Branagan, W.D. Swank, D.C. Haggart, J.R. Fincke, Metal. Mater. Trans. A 32 (2001)
2615-2621.
[11] D.J. Branagan, W.D. Swank, B.E. Meacham, Metal. Mater. Trans. A 40 (2009)1306-1313.
[12] J.C. Farmer, J.J. Haslam, S.D. Day, T. Lian, C.K. Saw, P.D. Hailey, J-S. Choi, R.B. Rebak,
J. Mater. Res. 22 (2007) 2297-2311.
[13] J.R. Davis, Handbook of Thermal Spray Technology, ASM International, Materials Park,
OH, 2004.
[14] L. Pawlowski, The Science and Engineering of Thermal Spray Coatings, John Wiley &
Sons Ltd., Chichester, West Sussex, England, 1995.
[15] P. Fauchais, A. Vardelle, B. Dussoubs, J. Therm. Spray Technol. 10 (2001) 44-66.
34
[16] K.S. Raja, M. Misra, K. Paramguru, Mater. Lett. 59 (17) (2005) 2137-2141.
[17] M.S. Phadke, Quality Engineering Using Robust Design. Engelwood Cliffs, PTR Prentice
Hall, New Jersey, 1989.
[18] G. Taguchi, S. Konishi, Taguchi Methods: Orthogonal Arrays and Linear Graphs, ASI
Press, 1987.
[19] J. Cizek, K.A. Khor, Z. Prochazka, Mater. Sci. Eng. C 27 (2007) 340–344.
[20] G. Bertrand, P. Bertrand, P. Roy, C. Rio, R. Mevrel, Surf. Coat. Technol. 202 (2008)
1994–2001.
[21] J.A. Hearley, J.A. Little, A.J. Sturgeon, Surf. Coat. Technol. 123 (2000) 210–218.
[22] O. Maranho, D. Rodrigues, M. Jr Boccalini, A. Sinatora, Surf. Coat. Technol. 202 (15)
(2008) 3494-3500.
[23] E. Turunen, T. Varis, S-P Hannula, A. Vaidya, A. Kulkarni, J. Gutleber, S. Sampath, H.
Herman, Mater. Sci. Eng. A 415 (2006) 1–11.
[24] H. Choi, S. Lee, B. Kim, H. Jo, C. Lee, J. Mater. Sci. 40 (2005) 6121–6126.
[25] D. Zois, A. Lekatou, M. Vardavoulias, J. Alloys Compd. 504S (2010) S283-S287.
[26] X.C. Zhang, B.S. Xu, Y.X. Wu, F.Z. Xuan, S.T. Tu, Appl. Surf. Sci. 254 (2008) 3879–
3889.
[27] E. Lugscheider, C. Herbst, L. Zhao, Surf. Coat. Technol. 108–109 (1998) 16–23.
[28] H.B. Guo, F. Vaßen, D. Stöver, Surf. Coat. Technol. 186 (2004) 353– 363.
[29] X.Q. Liu, Y.G. Zheng, X.C. Chang, W.L. Hou, J.Q. Wang, Z. Tang, A. Burgess J. Alloys
Compd. 484 (2009) 300-307.
[30] Z.H. Gan, H.Y. Yi, J. Pu, J. F. Wang, J.Z. Xiao, Scripta Mater. 48 (2003) 1543-1547.
[31] Q.J. Chen, H.B. Fan, J. Shen, J.F. Sun, Z.P. Lu, J. Alloys Compd. 407 (2006) 125-128.
[32] Q.J. Chen, H.B. Fan, L. Ye, S. Ringer, J.F. Sun, J. Shen, D.G. McCartney, Mater. Sci. Eng.
A 402 (2005) 188-192.
35
[33] D. Zois, A. Lekatou, M. Vardavoulias, Surf. Coat. Technol. 204 (2009) 15-27.
[34] V. Keryvin, V.H. Hoang, J. Shen, Intermetallics 17 (2009) 211-217.
[35] R.S. Lima, B.R. Marple, Surf. Coat. Technol. 200 (2006) 3428-3437.
[36] Z. Burghard, A. Zimmermann, J. Rödel, F. Aldinger, B. R. Lawn, Acta Mater. 52 (2004)
293-297.
[37] M.J. Owen and R.G. Rose, J. Phys. D: Appl. Phys 6 (1973) 42-53.
[38] W-Z Wang, C-J Li, Y-Y Wang, G-J Yang, K. Sonoya, Surf. Coat. Technol. 200 (2006)
842-847.
[39] J.G. La Barbera-Sosa, Y.Y. Santana, M.H. Staia, D. Chicot, J. Lesage, J. Caro, G.
Mesmacque, E.S. Puchi-Cabrera, Surf. Coat. Technol. 202 (2008) 4552-4559.
[40] J. F. Loffler, Intermetallics 11 (2003) 529-540.
[41] A. Inoue, L. Arnberg, M. Oguchi, U. Backmark, N. Böckström, T. Masumoto, Mater. Sci.
Eng. 95 (1987) 101-114.
[42] M.E. Aalamialeagha, S.J. Harris, M. Emamighomi, J. Mater. Sci. 38 (2003) 4587–4596.
[43] L. Ajdelsztajn, E.J. Lavernia, B. Jodoin, P. Richer, E. Sancousy, J. Therm. Spray Technol.
15 (2006) 495-500.
[44] ASM Handbook: Alloy Phase Diagrams, vol. 3, ASM International, Materials Park, OH,
1997.
[45] Y. Lin, J. Hu, J. Mater. Sci. 26 (1991) 2833-2840.
[46] C. Guo, P.M. Kelly, Mater. Sci. Eng. A 352 (2003) 40-45.
[47] P. Rogl, Boron-Iron-Tungsten, in Springer Materials: The Landolt-Börnstein Database,
DOI: 10.1007/978-3-540-69761-9.26, Springer 2008, http://www.springermaterials.com.
[48] Y.E. Kalay, L.S Chumbley, I.E. Anderson, Mater. Sci. Eng A 490 (2008) 72-80.
[49] P. Dong W.L. Hou, X.C. Chang, M.X. Quan, J.Q. Wang, J. Alloys Compd. 436 (2007)
118-123.
36
[50] Y. Liu, Z. Liu, S. Guo, Y. Du, B. Huang, J. Huang, S. Chen, F. Liu, Intermetallics 13
(2005) 393–398
[51] A. Inoue, B.L. Shen, C.T. Chang, Intermetallics 14 (2006) 936-944.
[52] J.Y. Bang, R.Y. Lee, J. Mater. Sci. 24 (1991) 4961-4965.
[53] A. Grabias, D. Oleszak, M. Kopcewicz, J. Latuch, Mater. Sci. Eng. A 449-451 (2007) 552-
556.
[54] C.N. Guy, A.A. Uraz, J. Less Common Met. 48 (1976) 199-203.
[55] D. Shin, F. Gitzhofer, C. Moreau, J. Therm. Spray Technol. 16 (2007) 118-127.
[56] J. Kim, C. Lee, H. Choi, J. Ho, H. Kim, Mater. Sci. Eng. A 449-451 (2007) 858-862.
[57] K.S. Narasimhan, F.J. Semel, in: Proc. 2007 Intl. Conf. on Powder Metallurgy and
Particulate Materials, May 13-16, Denver, Colorado, Metal Powder Industries Federation,
Paper No. 2007-01-0145.
[58] V. Raghavan, J. Phase Equilib. Diffus. 28 (2007) 380-381.
[59] Q. Wei, B. Yang, A. Jiang, Chin. Phys. Lett. 8 (1991) 296-299.
[60] G. Roberts, G. Krauss, R. Kennedy, Tool Steels, 5th ed., ASM International, Materials
Park, OH, 1998.
[61] A.S. Chaus, Phys. Met. Metallogr. 106 (1) (2008) 82-89.
[62] T. Kulik, J. Non-Crystal. Solids 287 (2001) 145-161.
[63] V.M. Kardonskii, Met. Sci. Heat Treat. 12(7) (1970) 547-550.
[64] H.E. Khalifa, J.L. Cheney, K.S. Vecchio, Mater. Sci. Eng. A 490 (2008) 221-228.
[65] H. Luo, D. Goberman, L. Shaw, M. Gell, Mater. Sci. Eng. A 346 (2003) 237-245.
[66] A.P. Newbery, P.S. Grant, J. Mater. Process Technol. 178 (2006) 259–269.
37
[67] D. Zois, A. Lekatou, M. Vardavoulias, I. Panagiotopoulos, A. Vazdirvanidis, J. Therm.
Spray Technol. 17 (2008) 887-894.
[68] M. Gaona, R.S. Lima, B.R. Marple, J. Mater. Process. Technol. 198 (2008) 426-435.
[69] P Bansal, P.H. Shipway, S.B. Leen, Acta Mater. 55 (2007) 5089-5101.
[70] U. Ramamurty, S. Jana, Y. Kawamura, K. Chattopadhyay, Acta Mater. 53 (2005) 705-717.
[71] Y. Wu, P. Lin, G. Xie, J. Hu, M. Gao, Mater. Sci. Eng. A 430 (2006) 34-39.
[72] C.A.C. Souza, J.E. May, L. Bolfarini, S.E. Kuri, M.F. de Oliveira, C.S. Kiminami J. Non-
Cryst. Solids 284 (2001) 99-104.
[73] H.X. Li, S. Yi, Mater. Chem. Phys. 112 (1) (2008)305-309.
38
Table I
The levels of the control factors.
Factor Levels
1 2 3
A. Oxygen-to-fuel
ratio
3.5 5.0 6.5
B. Powder feed rate
(g/min)
24 31 38
C. Spraying distance
(mm)
220 250 220a
D. Spraying stages yes: stages applied no: one stage yes: stages applieda
a Dummy level
39
Table II
The experimental conditions in the form of an L9 (34) orthogonal array.
Run No. Oxygen-to-
fuel ratio
Powder feed
rate (g/min)
Spraying
distance (mm)
Spraying
stages
13.5 24 220 yes
23.5 31 250 no
33.5 38 220 yes
45.0 24 250 yes
55.0 31 220 yes
65.0 38 220 no
76.5 24 220 no
86.5 31 220 yes
96.5 38 250 yes
40
Table III
DSC peak temperatures and corresponding enthalpies.
Materials
Tp1
(oC)
ΔΗ1
(J/g)
Tp2
(oC)
ΔΗ2
(J/g)
Tp3
(oC)
ΔΗ3
(J/g)
Tp4
(oC)
ΔΗ4
(J/g)
NS powder
642.
3
5.19
697.
9
10.74 765.4 1.56 786.5 1.39
NS powder -45μm 642 6.51
697.
2
14.56 773.3 4.05 802.4 1.58
Coat. No.2
625.
1
3.40
694.
8
9.79 752.3 6.91 - -
Coat. No.5 626 3.55
698.
5
10.50 736.4 2.09 - -
Coat. No.8
625.
2
3.36 696.
4
9.83 738.9 2.11 - -
41
Table IV
Coating properties per experiment (mean values as reported in the Experimental).
Run No. Microhardness
(HV0.3Kg/10s)
Porosity
(%)
Crack extension force
Gc (J/m2)
Deposition
rate (μm/pass)
1879 ± 76 1.1 ± 0.5 9.6 ± 2.2 9.1 ± 1.5
2960 ± 66 1.1 ± 0.3 13.4 ± 2.2 7.0 ± 1.0
31032 ± 95 0.7 ± 0.2 42.3 ± 5.9 16.2 ± 1.1
4840 ± 105 0.8 ± 0.4 5.9 ± 0.7 10.0 ± .0.4
5860 ± 96 0.7 ± 0.4 24.5 ± 3.0 14.4 ± 0.5
6903 ± 63 0.4 ± 0.1 51.2 ± 5.2 14.1 ± 0.5
7727± 82 1.2 ± 0.4 18.6 ± 3.0 4.3 ± 0.5
8802 ± 73 1.1 ± 0.4 10.4 ± 1.5 11.0 ±0.6
9846 ± 94 0.9 ± 0.1 37.9 ± 5.9 15.4 ±0.9
42
Table V
The expected positions of the elements contained in the nanosteel powder (as well as of O
and Ag). in an EDX spectrum.
Elements Kα1 (keV) Κβ1 (keV) Lα1 (keV) Mα1 (keV) Mβ (keV)
B 0.19
C 0.28
O 0.52
Si 1.74
Cr 5.41 5.95 0.57
Mn 5.90 6.49 0.64
Fe 6.40 7.06 0.70
Mo 2.29
Ag 2.98
W 8.40 1.78 1.84
43
Table VI
Effect of factors on the coating properties by means of the variance ratio F.
Factor
F
(Microhardnes
s)
F
(Porosit
y)
F
(Deposition rate)
F
(Gc)
A. Oxygen-to-Fuel
ratio
16.3 7.1 2.3 <1
B. Powder feed rate
(g/min)
7.4 4.8 16.2 23.5
C. Spraying
distance (mm)
<1 1.0 <1 2.6
D. Spraying stages <1 <1 13.8 1.8
44
Table VII
The anticipated values of coating properties under the optimum spraying conditions and the
verification experiment values.
Properties
Property mean values at factor optimum
levels
Anticipate
d values
Verificatio
n
experiment
values
Oxygen-
to-Fuel
ratio (5)
Powder
feed rate
(38
g/min)
Spraying
distance
(220 mm)
Sprayin
g stages
(Yes)
Hardness
(HV0.3)
868 927 - a-922 946±90
Porosity
(%)
0.7 0.7 0.9 - 0.4 0.4±0.1
Deposition
rate
(μm/pass)
13 15 - 13 18 15±1
Gc (J/m2)- 43.8 26.1 27.7 50.1 45.5±2.6
a dashes: F<1
45
Figure captions
Fig. 1. Backscattered electron micrograph of the stainless steel powder feedstock .
Fig. 2. XRD patterns of powders (as fed and after -45 μm sieving) and surfaces of coatings as-
sprayed (No. 2: A1 level, No. 5: A2 level, No. 8: A3 level). 1: α (Fe,Cr) (34-396), 2: Fe1.86C0.14
(44-1289), 3: Fe23B6 (47-1332), 4: Fe23(B,C)6 (12-570), 5: Fe15Si3B2 (47-1629), 6: Fe3B (39-
1316), 7: Cr1.65Fe0.35B0.96 (35-1180).
Fig. 3. (a) DSC plots of powders (as fed and after -45 μm sieving) and coatings (No. 2, No. 5
and No. 8). (b) The low temperature exothermic peak of the coatings (arrow pointed). (c) The
high temperature exothermic peak of coating No. 2 may be composed of two peaks at 752 and
764 oC.
Fig. 4. (a) XRD patterns of the annealed standard NS powders. (b) A magnification of the
patterns in the range 2θ=60-70 ο. 1: α (Fe,Cr) (34-396), 2: Fe1.86C0.14 (44-1289), 3: Fe23B6 (47-
1332), 4: Fe23(B,C)6 (12-570), 5: Fe15Si3B2 (47-1629), 6: Fe3B (39-1316), 7: Cr1.65Fe0.35B0.96 (35-
1180), 8: WC (72-97), 9: Cr23C6 (71-552), 10: WC1-x (20-1316), 11: CrB2 (75-1043), 12: Fe3O4
(76-1849).
Fig. 5. Cross-section of coating No. 5 (BSE mode). A Gc indentation mark is discerned (left
hand-side).
Fig. 6. (a) A semi-melted zone in coating No. 3 (BSE mode). (b) Typical EDX analysis from the
matrix of the precipitates in 6a. (c) EDX analysis from the arrow pointed precipitate in 6a.
Fig. 7. (a) Oxide stringer (white arrow) and unmelted particles (black arrow) in coating No. 7
(BSE mode). (b) EDX analysis from the oxide stringer.
Fig. 8. XRD patterns of the surfaces of the annealed coatings {(a): No. 2, (b): No. 5, (c) No. 8}.
1: α (Fe,Cr) (34-396), 2: Fe1.86C0.14 (44-1289), 3: Fe23B6 (47-1332), 4: Fe23(B,C)6 (12-570), 5:
46
Fe15Si3B2 (47-1629), 6: Fe3B (39-1316), 7: Cr1.65Fe0.35B0.96 (35-1180), 8: WC (72-97), 9: Cr23C6
(71-552), 10: WC1-x (20-1316), 11: CrB2 (75-1043), 12: Fe3O4 (76-1849).
Fig. 9. (a) Nanocrystallization inside a crack, SE mode (coating No. 2 annealed at 770 oC, for 2
h). (b) EDX analysis of a nano-crystal (1). (c) EDX analysis from the matrix close to the crack
(2).
Fig. 10. A pocket with Si-rich nanocrystallites, enveloped by a W-rich/Si-poor layer (SE mode).
EDX analyses from the: (b) Si-rich pocket (1), (c) W-rich/Si-poor layer (2), (d) α-FeCr matrix
(3) (coating No. 2 annealed at 770 oC, for 2 h).
Fig. 11. Plots of factor effects on coating properties: (a) microhardness; (b) porosity; (c)
deposition rate; (d) crack extension force. The mean value for each property (average of 9 run
responses) is shown by the dotted line. A: oxygen-to-fuel ratio, B: powder feed rate (g/min), C:
spraying distance (mm), D: spraying stages.
Fig. 12. Indentation crack propagating along the splat boundaries and parallel to the substrate
surface (coating No. 3 in cross-section, 1kg/s).
Fig. 13. XRD patterns of the free surface side and the substrate side of coating No. 6 (detached
from its substrate).
Fig. 14. (a) The coating produced by the verification experiment under the optimized spraying
parameters extracted by the Taguchi analysis (Secondary Electron mode); (b) The XRD pattern
of the verification coating.
47
Fig. 1
48
Fig. 2
49
20 30 40 50 60 70 80
2θ (
ο
)
Intensity (a.u.)
2
2
2,3
2,3
2,3
12
7
2
2
2
2
4,7
7
5
5,6,7
2
7
NS powder
NS -45 μm
coat. No. 2
coat. No. 5
coat. No. 8
5,6
3
3
3
4
2
5,6
3
7
4,7
5
4
5,6
5
4
2
4
4
4,7
Fig. 3
50
(a)
(b)
(c)
550 600 650 700 750 800
Temperature (
o
C)
NS -45 μm
NS powder
coat. No. 2
coat. No. 5
coat. No. 8
Exo
700 720 740 760 780 800 820
Temperature (
o
C)
NS powder
coat. No. 2
coat. No. 5
Exo
Fig. 4
51
(b)
20 30 40 50 60 70 80
2θ (
ο
)
Intensity (a.u.)
710
o
C
2
2
10
NS powder
655
o
C
775
o
C
795
o
C
2,3
2
2
7
4,7
5,6,7
4
5,6
3
3
4
2,3
5,6,7
2
5
7
3
5,6
11
11
11
11
11
11
2
1,4,7
4
4
10?
4
1,4,7
5,6,7
4
4
3
7
5,6
4
3
1,2
4
2
1
9
9
9
1
11
11
1
9
9
9
9
9
9
2
10
10
1
9
3
9
1
1
2
9
1
2
(a)
Intensity (a.u.)
2θ (ο)
655 oC
710 oC
775 oC
795 oC
60 70
12
1,2
1
1
Fig. 5
52
Fig. 6
53
(a)
C
Mo
Ag
Cr
Fe
W
W
Fe
B
(b)
(c)
Fig. 7
54
Mo
O
Cr
Fe W Cr Fe
123456
keV
(b)
(a)
55
Fig. 8
20 30 40 50 60 70 80
2θ (
ο
)
Intensity (a.u.)
10
2
coat. No. 2
640
o
C
710
o
C
770
o
C
2,3
4,7
4
5
7
2
4
4,7
2,3
2
6
3
5
4
3
7
2
4
6
4
5
7
7
4
8
7
7
4
5,6
2,3
1
1
6
5,8
3
4
4
5
7
1
4
4
9
9
1
9
9
9
11
9
11
10
8,11
2
8
9
9
3
9
9
4
4
1
1
2
1
8
2
8
8
8
3
3
(a)
20 30 40 50 60 70 80
2θ (
ο
)
Intenisty (a.u.)
9
coat. No. 5
640
o
C
710
o
C
755
o
C
1
1,4,7
2
7
2,3
4,7
2
4
7
7
9
8
4
3
3
4
5,8
4
9
7
1
8
7
4
5
1
8
8
4
8
8,9
2,3
10
10
3
1
2
9
9
8
5
11
11
2
2
11
4
4
11
11
11
3
1
10
10
5,8
5
2
10
4
(b)
20 30 40 50 60 70 80
2θ (
ο
)
Intensity (a.u.)
coat. No. 8
640
o
C
710
o
C
755
o
C
12
2
2
5
2,3
1
1
7
7
4
1
9
2
5
5
5
4
3
12
5
4,7
4
4
1
2,3
4
5,9
9
4
9
4
4
12
9
1
2
9
11
9
9
11,12
9
11
11
11
7
3
7
2
2
2
10
10
9
2
3
3
(c)
Fig. 9
56
2 4 6
keV
8 10
Cr
Fe
Mo Fe W
W
Cr
2
2
1
(a)
2
keV
2 4 6 8 10
Cr
Cr
Fe
Mo Fe W
W
1(b)
(c)
Fig. 10
57
m
μ
2
2
3
1
(a)
1(b) 2(c)
3(d)
Fig. 11
(a)
(c) (d)
(b)
58
Fig. 12
59
Fig. 13
20 30 40 50 60 70 80
2θ (
ο
)
Intensity (a.u.)
Substrate side
Surface side
60
Fig. 14
20 30 40 50 60 70 80
2θ (
o
)
Intensity (a.u.)
(b)
(a)
61
