Available online at www.sciencedirect.

com

CERAMICS
INTERNATIONAL
Ceramics International 41 (2015) 11581–11591
www.elsevier.com/locate/ceramint

Review paper

Characterisation, significance and detection of manufacturing defects

in Reaction Sintered Silicon Carbide armour materials
Ian G Croucha,c,n, Manasa Kesharajub,c, Romesh Nagarajahb,c
a
Armour Solutions, Trentham, Victoria 3458, Australia
b
Faculty of Engineering and Industrial Sciences, Swinburn University of Technology, Hawthorn, Victoria 3122, Australia
c
Defence Materials Technology Centre, Hawthorn, Victoria 3122, Australia
Received 13 May 2015; received in revised form 15 June 2015; accepted 17 June 2015
Available online 25 June 2015

Abstract

Ceramics with high hardness values ( 4 10 GPa), high elastic modulii ( 4 300 GPa) and good flexural strength values ( 4 350 MPa) make
excellent armour materials. However, they are normally very expensive ( 4 $10/kg) and difficult to shape. So, tiles of regular geometry (typically
square or hexagonal) have been used in advanced armour systems, for military vehicles. However, for body armour applications, there is a need
to form a monolithic ceramic into a double-curved, or even triple-curved, shape in order to better conform to the human torso. In a recent review,
carried out by the Defence Materials Technology Centre of Australia, of the shaping techniques used for body armour components, uniaxial
pressing of breastplates was still recognised to be the most feasible, and commercially available, process even though other techniques like
Viscous Plastic Processing (VPP) were becoming established. One particular variant of the dry-pressing route is known as Reaction Sintering, or
Reaction Bonding, in which a pre-pressed, dry powder compact is infiltrated, at elevated temperature, with a liquid metal. However, the process is
prone to the formation of a range of manufacturing defects, some of which are extremely deleterious to impact behaviour and ballistic
performance of the finished product. The origin, cause and effect of these defects are discussed in this paper, together with both current and future
NDI techniques.
& 2015 Elsevier Ltd and Techna Group S.r.l. All rights reserved.

Keywords: Ceramic armour; Reaction Sintered Silicon Carbide; Manufacturing defects; Non-destructive inspection

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11582
2. Penetration mechanisms in ceramic armours . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11582
3. Processing details for Reaction Sintered Silicon Carbide (RSSC) products . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11584
3.1. Mixing of the original set of powders . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11585
3.2. Pressing of preform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11586
3.3. Curing of preform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11586
3.4. Firing of preform, dipped in bath of molten silicon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11586
4. Characterisation of defects, or heterogeneities, in RSSC products . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11587
4.1. Thickness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11587
4.2. Bulk density . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11587
4.3. Porosity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11587
4.4. Unsintered material. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11587

n
Corresponding author at: Armour Solutions, Trentham, Victoria 3458, Australia. Tel.: þ 61 402 140 911.
E-mail address: ianarmoursolutions@gmail.com (I. Crouch).

http://dx.doi.org/10.1016/j.ceramint.2015.06.083
0272-8842/& 2015 Elsevier Ltd and Techna Group S.r.l. All rights reserved.
11582 I.G Crouch et al. / Ceramics International 41 (2015) 11581–11591

4.5. Silicon-rich areas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11588

4.6. Cracks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11588
4.7. River markings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11588
5. X-radiography of RSSC products . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11588
6. Use of ultrasonic detection method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11589
6.1. Local density variation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11589
6.2. Manufacturing defects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11590
7. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11590
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11591
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11591

1. Introduction Interest in this particular process is reflected in the use of

Ceramics with high hardness values (410 GPa), high elastic
modulii (4300 GPa) and good flexural strength values
(4350 MPa) make excellent armour materials [1]. However,
they are normally very expensive (4$10/kg) and difficult to
shape. So, tiles of regular geometry (typically square or hex-
agonal) have been used in advanced armour systems, for military
vehicles, offering high levels of protection against kinetic energy
weapons like armour-piercing rifle bullets (up to 14.5 mm in
calibre) and small-calibre cannon rounds (up to 30 mm in
diameter) [2]. However, for body armour applications, Crouch
reported in 2009 [3], that there is a need to form a monolithic
ceramic into a double-curved, or even triple-curved, shape in
order to better conform to the human torso. Fig. 1 shows a typical
body armour production facility and associated products. In a
recent review, carried out by Leo et al. [4], of the shaping
techniques used for body armour components, uniaxial pressing
of breastplates was still recognised to be the most feasible, and
commercially available, process even though other techniques
like Viscous Plastic Processing (VPP) were becoming estab-
lished. One particular variant of the dry-pressing route is known
as Reaction Sintering, or Reaction Bonding, in which a pre-
pressed, dry powder compact is infiltrated, at elevated tempera-
ture, with a liquid metal. However, the process is prone to the
formation of a range of manufacturing defects, some of which are
extremely deleterious to impact behaviour and ballistic perfor-
mance. The origin of these defects is discussed in this paper,
together with both current and future NDI techniques.
Reaction Sintered Silicon Carbide (RSSC) as the main
projectile-eroding material in many body armour components
used by military personnel in both the United States (via M-
cubed Technologies) and Australia (via Modern Ceramics
Company (MCC)), since the late 1990s [5,6]. Prior to its
widespread adoption in Australia by Australian Defence Appa-
rel (ADA) around 2008, extensive characterisation of possible
manufacturing defects had to be carried out and in-depth quality
acceptance criteria established. This paper reports details of
those quality control procedures and, in particular, compares the
advantages and disadvantages of established, high-resolution,
digital x-radiography with the use of ultrasonic methods,
recently researched by Kesharaju [7].
2. Penetration mechanisms in ceramic armours
Fig. 2a shows the well-established steps in the penetration of a
ceramic armour system which, in the main, consists of two layers:
a ceramic element on the impact side, supported by a backing
material on the distal side; this material is typically an aluminium
alloy, steel or a fibre-reinforced polymer. The impacting bullet is
partly eroded at the surface during a time-period known as dwell,
followed by the formation of Hertzian cracks on the surface of the
ceramic, and final failure through conoidal fracture. The formation
of the conoid increases the area of contact with the backing
material which then bulges and absorbs energy principally through
membrane stretching. This local failure is accompanied by the
formation of radial cracks (see Fig. 2b), which emanate from the

Fig. 1. Production of HAPs at ADA, Bendigo, Australia, circa 2010.

 I.G Crouch et al. / Ceramics International 41 (2015) 11581–11591 11583

Fig. 2. (a) Stages in through-thickness penetration of ceramic armours [8], and (b) example of radial cracking propagating from Point of Impact (PoI) [9].

Fig. 3. A cylindrical plug of RSSC material sheared out of a 9 mm thick RSSC plate when being ballistically tested with an impact from a 7.62 mm APM2 round.
LHS shows impact face. The RHS image shows a side view of the plug – the heterogeneous nature of the microstructure is quite clear.

Table 1
Physical properties of RSSC material, produced by MCC (Sydney) circa 2009. The tests were carried out in the School of Materials Science and
Engineering, UNSW.

Property Units Value Test method UNSW report

Average grain size mm 19.078.0 ASTM E112-96 (2004) 09-1309

ASTM grain size number, G 8.5 ASTM E112-96 (2004) 09-1309
Bulk density kg/m3 3106713 ASTM C20-00 (2005) 09-1308
Apparent porosity % 0.2270.06 ASTM C20-00 (2005) 09-1308
Sonic velocity m/s 11,3287 53 ASTM C1419-99a (2004) 09-1310
Young's modulus GPa 39974 ASTM C1419-99a (2004) 09-1310
Knoop microhardness GPa 24.570.4 ASTM C1326 (2008) 09-1311
Flexural strength MPa 504741 ASTM C1161-02c (2008) 09-1315
11584 I.G Crouch et al. / Ceramics International 41 (2015) 11581–11591
30
Residual Silicon Content [Vol.%]
25
20
15
10
5
0 of-the-art furnace technology, can produce a quality product at
2980 3000 3020 3040 3060 3080
Bulk Density [kg/m3]
Fig. 4. Effects of residual silicon content upon bulk density of RSSC products
(MCC production data, circa 2005).
Point of Impact (PoI), followed by circumferential cracking of the
individual segments.
It is absolutely essential that armour-grade ceramics fail in this
particular manner in order for their high level of ballistic performance
to be maintained. It is therefore critically important to identify any
material defect which may affect, or change, this mode of failure. Of
greatest importance is the dynamic, through-thickness, compressive
modulus of the material, immediately in front of the impacting
projectile (at Stages a and b in Fig. 2a). This cylinder of material,
similar in size to the core diameter of the bullet (about 6–8 mm), must
be as rigid as possible and resist the initial impact. So, the presence of
porosity, or poor quality material, in that cylindrical zone, or the
existence of a crack, or tear, running through, or near, the cylindrical
zone, may well reduce its compressive modulus, and therefore its
penetration resistance. For example, Crouch recently reported [9] that
an impact adjacent to a crack, with a Crack Opening Displacement
(COD) of up to 1 mm, reduces the ballistic performance by up to 9%.
On the other hand, the effects of either local porosity, or poorly
processed material, in that cylindrical zone have been more difficult to
quantify. However, common-sense, material science and engineering
rules have been applied when considering the likely effects of such
defects. As can be seen from Fig. 3, if microstructural defects are
present then the poor result is quite evident – Fig. 3 shows a plug of
RSSC material ejected from a 9 mm thick breastplate tile impacted by
a 7.62 mm APM2 round during ballistic testing. This is a good
example of an unacceptable failure mode in ceramic armours – very
little impact energy will have been absorbed as a result of the ceramic
failing in this manner. The criticality of a range of manufacturing
defects is discussed in Section 4 of this paper.
Hard Armour Plates (HAP) are precision-engineered com-
ponents [3] based upon the use of a ceramic tile which has a
known bulk density, to within 7 10 kg/m3, and a uniform
thickness, to within 7 0.2 mm. This is essential because
ballistic performance of a HAP is a function of the areal
density of the finished product: areal density [kg/m2] is the
product of bulk density [kg/m3] and thickness [m]. The shape
of the finished HAP is dictated by the shape of the ceramic tile.
Since these products must exhibit a consistent level of ballistic
performance across the entire surface area of the HAP, the
microstructure, bulk density and thickness of the tile are
critically important and must be as consistent as possible.
3. Processing details for Reaction Sintered Silicon Carbide
(RSSC) products
Reaction sintering, or reaction bonding, of silicon carbide
and boron carbide ceramics is a much more cost effective
manufacturing process than hot pressing [1], especially for
production of shaped armour products. Whilst the process is
extremely complex, and proprietary in nature, it can utilise
lower cost, abrasive grades of powder, and, together with state-
3100 3120 3140
a reasonable price. Bulk density and hardness values are
slightly lower than their hot-pressed variants (see Table 1)
but the cost of manufacture, especially for the silicon carbide
grades, approaches that for standard alumina products. They
are therefore very attractive to the designer, manufacturer and
purchaser of body armour systems.
Other inherent advantages of these materials, compared with
both conventional aluminas and the hot-pressed non-oxides, offer
considerable scope for applying these new materials to Hard
Armour Plates (HAP) for protection against high-velocity rifle
rounds in body armour systems. The material has close to zero
porosity, thereby maximising ballistic performance. It has negli-
gible shrinkage, therefore precision products are achievable. It can
also be welded (during the firing stage) and it can be pre-cut, and
shaped, in the green state. Because of these attractive qualities,
RSSC materials represent a very user-friendly family of engineer-
ing materials. Table 1 lists the major physical and mechanical
properties typical of these RSSC materials.
The residual silicon (Si) content in the ceramic, inherent from the
manufacturing process, is a key manufacturing parameter since it
controls final bulk density (see Fig. 4) which, in turn, affects ballistic
performance. Since silicon is a softer material than SiC, the lower the
Si content, the harder the ceramic and, in general, the higher the
ballistic performance will be. Conversely, as the silicon content
decreases, the fracture toughness of the ceramic is likely to decrease
and this may affect the ceramic's multi-hit behaviour. For these
reasons, and based upon unpublished work carried out by co-author
Crouch, whilst at Australian Defence Apparel (ADA) between 2004
and 2008, grades with a bulk density between 3040 and 3080 kg/m3
are most suitable against multi-strikes from lead-filled, fully-jacketed
rounds like the US M80, whilst those between 3080 and 3120
perform better against single-strike, armour-piercing bullets, like the
Envisaged
direction of flow
Area of incomplete fill
leading to a large area
of unsintered material
Envisaged flow
front of liquid
silicon
Level of molten silicon
in furnace
Fig. 5. X-ray of defective RSSC breastplate tile, showing large area of
unsintered material, and (superimposed) indications of expected flow of liquid
silicon through the vertical preform.
 I.G Crouch et al. / Ceramics International 41 (2015) 11581–11591
Fig. 6. X-ray image of a large region of unsintered material in a 4 mm thick RSSC tile. The right-hand image shows the broken tile revealing the lenticular nature of
the internal defect.
US APM2 round. Similar work published elsewhere, for Reaction
Bonded Boron Carbide materials, [10], reports a sharp decrease in
ballistic performance against armour-piercing bullets ( 20%) when
increasing the silicon content from  17% to 27%, by volume.
RSSC materials were originally developed in the 1950s with
a number of US patents following between the mid-60s and
mid-70s [11,12]. Russian technologists were also hard at work
developing a similar process [13] but it was not until the late-
1990s that full-scale production was commercially available in
both the US and Australia. As mentioned previously, details of
the precise processing procedures are still very proprietary.
However, the following overview enables the reader to
appreciate the salient steps of the process and the origin, and
likely cause, of most of the inherent manufacturing defects.
More details are given elsewhere in a proprietary, technical
memorandum issued by MCC in 2009 and written by Dr
Evgeniy Popov, founder of this particular variant of the RSSC
process in 1998, Professor Andrew Ruys (University of
Sydney, Australia) and Dr Ian Crouch (Managing Director,
Armour Solutions Pty Ltd, Australia). The article was entitled
“Quality of Reaction Sintered Silicon Carbide products” and
issued to the Australian defence community.
In general, the process involves the infiltration of liquid
silicon, through a pre-pressed compact of a mixture of SiC and
11585
carbon particles. The liquid silicon reacts with the carbon to
form new-SiC. A residual amount of silicon remains when all
of the carbon has reacted. Thus, in principle, there is no
porosity, only isolated islands of metallic silicon, in a matrix of
original-SiC and new-SiC. However, the process relies upon
production of uniformly-compacted preforms of SiC, carbon
and an appropriate binding agent, normally a phenolic resin.
The resin is burnt off, and/or converted into carbon, during a
preparatory pyrolysis process. The porous preform is then
infiltrated with liquid silicon and this gives rise to another set
of possible defects. Because the infiltration process involves
flow of liquid silicon through the preform, a number of
casting-like defects are possible, like air-entrapment, cold
shuts and incomplete filling. Cold shuts are where two molten
surfaces come into contact but do not metallurgically bond or
fuse, forming a planar, crack-like defect.
The following steps in this reaction sintering process are
identified as critical to the formation, or prevention, of critical
defects within the final microstructure:
3.1. Mixing of the original set of powders
In order to obtain a final product with a homogenous
microstructure, the preform needs to be homogenous, as does
11586 I.G Crouch et al. / Ceramics International 41 (2015) 11581–11591
Fig. 7. X-ray image (right) and visual images of a Si-rich tear in a 9 mm thick RSSC breastplate tile. The X-ray image shows a slightly lighter region surrounding
the initial tear in the preform.
the original powder mixture. It is critically important that the
binder (usually a phenolic or epoxy resin) is uniformly distributed
throughout the mix to guarantee that the infiltration process
occurs at a steady, controlled rate throughout the preform.
3.2. Pressing of preform
This is the most critical step. Matched-metal, closed tooling
is normally used between the plattens of a heavy, uni-axial,
hydraulic press. The resultant preform represents the final
dimensional form of the finished product, in this case a 4–
10 mm thick breastplate tile. Charge weights are accurately
controlled, to the nearest gram, and the mould cavity is filled in
a very uniform manner, with the thickness closely toleranced,
because the applied pressure strongly influences the resulting
bulk density of the preform. If the preform varies in bulk
density, from place to place, so will the final product, since the
bulk density controls both the rate and extent of silicon
infiltration. Non-uniform applied pressures can therefore lead
to unacceptable variations in both thickness and bulk density.
Incorrect filling, or the entrapment of air-bubbles, can also lead
to the formation of small areas of porosity in the final product.
3.3. Curing of preform
Prior to firing, the preforms are pyrolysed in order to burn-off or
convert the binder material into free carbon. Out-gassing of the
product needs to be monitored and the entire process needs to be
complete, to ensure that no residual binder remains in the preform.
The open network of resultant porosity needs to be uniformly
distributed throughout the preform. The preform is fairly frangible
and small tears, especially around the periphery, are likely to occur if
the preforms are not cooled appropriately and/or mishandled.
3.4. Firing of preform, dipped in bath of molten silicon
Breastplate preforms are normally fired in the vertical
position with one end of the preform sitting in the molten
pool of silicon at about 1500–1700 1C. The hot silicon is
drawn into the preform via capillary action and, ideally is
drawn up at a constant, and uniform, rate such that the
sintering reaction, which is occurring in-situ at the liquid/solid
interface, has time to complete. Fig. 5 indicates the likely flow
of liquid silicon through the vertical preform.
This is also a critical step in the process. Each and every RSSC
product is very similar to an individual metal casting, since both
processes rely upon controlling the flow of molten metal. Because of
this, a very similar set of casting-like defects are possible including
incomplete filling leading to areas of unsintered material. Fig. 6
shows an X-ray of such a feature and the expected flow fronts of the
liquid silicon. If there are any pre-existing defects in the preform, like
tears, the liquid silicon will simply fill the void and create a tear filled
with silicon metal (see Fig. 7). Excess silicon can also migrate to the
surface causing pools of silicon metal on the surface of the ceramic
tile. So, whilst the final product has an overall bulk density, the
variation from point to point, as well as throughout the section of the
tile, can vary considerably. Measuring the bulk density by conven-
tional gravimetric means will NOT, therefore, indicate how variable
the bulk density might be within the product itself.
One very characteristic feature of an RSSC product, especially
one manufactured in a conventional kiln under inert gas (i.e. not
under vacuum), is the formation of river markings at the base of the
product adjacent to the original level of the molten silicon. Under
specific conditions, the partial pressure of silicon is such that gas
bubbles are created on the surface of the ceramic preform, close to
the liquid/solid interface. These gas bubbles evolve in such a way
that they flow upwards along the surface of the unsintered preform,
ahead of the liquid–solid interface and create a set of indentation
 I.G Crouch et al. / Ceramics International 41 (2015) 11581–11591
Fig. 8. X-ray image (right) and comparable visual image of river-markings on the surface of a 9 mm thick RSSC breastplate. Such defects can be up to 2 mm deep.
lines, up to 1–2 mm deep, in a pattern very like the delta of a river.
Because of their depth, they can destroy the local surface profile/
geometry and become unacceptable defects. Fig. 8 shows an
example of this particular feature (using both visual and X-ray
images).
4. Characterisation of defects, or heterogeneities, in RSSC
products
Table 2 provides a summary of the possible manufacturing
defects in RSSC products, their likely effect upon ballistic
behaviour and their detectability using either x-radiography or
ultrasonic techniques. Each defect, or heterogeneity, is further
characterised below.
4.1. Thickness
The thickness of ceramic tiles used in HAPs are normally
controlled to within 70.2 mm which is quite a challenge for a
ceramic part not undergoing any form of secondary grinding or
polishing. Any deviation outside this range, especially lower than
a stated minimum value, is extremely significant. In practice,
measuring the thickness would be extremely tedious, and
normally such ceramic tiles are only measured (checked) in five
to ten locations, depending upon the specification. Surface finish
is therefore also important. Any surface flaw or defect, especially
an indentation like a river marking can be quite disruptive.
4.2. Bulk density
As has been stated above, the areal density of these products
needs to be very consistent. However, in RSSC products, many
of the local defects or flaws, can affect the local bulk density. It is
also possible that different combinations of defects, in one
11587
location, could result in the same value of measured bulk density.
Accurate determination of local bulk density values was the main
driver for research into ultrasonic detection methods, since subtle
variations in bulk density is difficult to detect by x-radiography.
The overall bulk density is normally set at a range, say from 3.04
to 3.08, but unless local measurements are made, values outside
this range can go undetected and values less than the stated
minimum could significantly affect ballistic performance.
4.3. Porosity
The reaction sintering process, using a liquid infiltration technique,
is inherently resistant to the formation of conventional porosity.
However, very occasionally, some entrapped air during compaction
of the preform can lead to localised porosity. Gas bubble evolution
during the firing process can also lead to surface, or near-surface,
porosity. The size of these pores is usually less than 2 mm.
4.4. Unsintered material
This is the most critical defect within an RSSC armour product
since unsintered material has very little through-thickness com-
pression strength – it is, by its nature, very frangible. Unfortu-
nately, the size and location of these areas can be quite varied,
depending upon the flow conditions within the preform. It is not
unusual to see areas of unsintered material in the centre of the
section and be lenticular in shape – see Fig. 6. They are normally
encapsulated within the body of the finished product so are
difficult to locate, except through radiography where they appear
as a dark, ill-defined region. Thinner plates are far more
susceptible to the occurrence of this type of defect since the
surface-to-volume ratio of the product is higher and therefore
surface tension effects are more influential.
11588 I.G Crouch et al. / Ceramics International 41 (2015) 11581–11591

Table 2
Likely causes and effects of manufacturing defects in RSSC products, their likely effect upon ballistic behaviour and their detectability using either x-radiography or
ultrasonic techniques.

Name of Nature of defect Possible cause of defect Possible effect upon Detectability by x- Detectability by
defect or penetration resistance radiography ultrasonic
heterogeneity techniques

4.1 Thickness Local variation in thickness Non-uniform filling of Marginal, if less than Extremely difficult Good, since sonic
mould during preform 0.3 mm flight times vary
manufacture with local
thickness
4.2 Bulk density Local change in through-thickness, bulk Non-uniform compaction Significant, if lower Difficult, especially if Good, since
density pressure during preform than acceptable range changes are subtle technique is
manufacture sensitive to subtle
changes in local
BD
4.3 Porosity Conventional holes/gaps in microstructure, Entrapment of air during Significant, if diameter Excellent, in sizes Good, in sizes
especially near centre of section filling of mould or 42 mm 4 1 mm 42 mm
compaction of preform.
Incomplete filling by
silicon metal
4.4 Unsintered Preform material which has not taken part Non-uniform flow of Very significant, if Excellent, in sizes Excellent, in sizes
material in the reaction process. Often found in silicon metal throughout diameter or thickness 4 1 mm 41 mm
isolated pockets, towards centre of section. the preform greater than 5% of
When fractured, they appear black and are nominal thickness of
very frangible – see Fig. 6 product
4.5 Silicon-rich Isolated areas of pure silicon metal, as in Formation of tears in Minimal effect, Good, if in correct Good, since pure
areas in-filling of tears and formation of puddles preform.Excess silicon especially if forming orientation and thick Si has higher
on surface of product – see Fig. 7 metal flowing to surface of continuous structure enough. transradiancy than
preform ceramic materials
4.6 Cracks Conventional cracking of brittle sintered Local stress reliefExternally Marginal depending Good, if cracks are Excellent, if
ceramic from various point defects applied load upon location and COD through-thickness and cracks are at an
– see Ref. [9] normal to surface obtuse angle to
the ultrasonic
beam
4.7 River A network of surface intrusions to a depth Caused by gas evolution on Significant if depth of Excellent since they Not easy since
markings of  1 mm, which take the form of a river the surface of the preform, defects are 45% of form on the surface they are surface
delta – see Fig. 8 near the molten silicon nominal thickness and reduce the local defects.
bath. thickness by up to
These defects are quite obvious to the Affected by partial pressure 2 mm
naked eye – see Fig. 8 of reactive silicon in the
firing process

4.5. Silicon-rich areas
It is a feature of the reaction sintering process to have small
islands of silicon metal uniformly distributed throughout the
microstructure, since this is one of the fundamentals of the process.
However, there are occasions when larger areas are formed. For
example, a silicon-filled tear can be seen in Fig. 7a and b. Pure
silicon has a measured hardness of  1250 kg/mm2 [6] compared
with values of SiC of  2500 kg/mm2 and can be considered to be
much tougher and able to resist crack propagation much more easily.
4.6. Cracks
These are conventional linear features found in many brittle,
and ductile, materials. They normally form in order to relieve
local stress or as a result of an external force (e.g. being
dropped or knocked). In RSSC armour tiles most of these
cracks are through-thickness cracks and form at right-angles to
the surface. They are therefore fairly easy to detect using
conventional x-radiography. However, such cracks are not
normally found in as-manufactured products.
4.7. River markings
The origin of these is described above, in Section 3d and
examples are shown in Fig. 8a and b. They are normally
unacceptable defects within an armour product and are very
visible to the naked eye. No NDI technique is required to
detect them.
5. X-radiography of RSSC products
Table 2 shows that x-radiography is very effective for most
of the manufacturing defects found within RSSC products,
especially those listed at 4.3 to 4.7. This semi-quantitative
assessment has been comprehensively used as the basis for
 I.G Crouch et al. / Ceramics International 41 (2015) 11581–11591
Fig. 9. 3D plot of the bulk density variation across an RSSC breastplate tile [7,16].
Fig. 10. Ultrasonic mapping of a ceramic tile (Tile 55) [7], defect free
region , porosity (density variation) , un-sintered material , variation in
thickness and small porosity , and border of ceramic tile.
quality documents throughout the entire procurement cycle,
from purchasing specifications and internal quality control
Work Instructions, to the establishment of quality assurance
standards for a range of clients from the Australian Defence
Force (ADF) to international police services.
HAPs, and associated ceramic breastplate tiles, supplied to the
ADF, are inspected at their approved facility in Melbourne,
Australia: Ballistic and Mechanical Testing (BMT) in Port
Melbourne. This company uses high-resolution, digital X-ray
equipment, as supplied by GE Inspection Technologies, using the
analytical software, Rhythm DICONDE Viewer 4.2. The X-ray
equipment has a resolution of less than 100 μm. Images are
digitally stored, enhanced (if required) and used as reference data
throughout the life of the product. More than 100,000 images
11589
have been generated over the past ten years or so. However,
whilst the quality of the images is excellent, x-radiography, by its
very nature, is extremely laborious. Analysis of the images is
especially time-consuming and often quite subjective, even
though quality acceptance criteria have been standardised. Recent
work by Haynes and others [14], to successfully automate x-
radiography, was strictly limited to the detection of cracks, unlike
in the present work where an array of defect types are of interest.
As mentioned above, Table 2 shows that conventional x-
radiography is good/excellent at detecting most of the manufactur-
ing defects (4.3 to 4.7). However, the technique is neither sensitive
enough to measure local variations in thickness (4.1) nor bulk
density (4.2), and both of these parameters are crucial in an armour
product, since Areal Density (the product of thickness and bulk
density) controls, by and large, ballistic performance. Determining
local bulk density values, from point-to-point, would therefore be
extremely useful in RSSC products. At the present time, bulk
density measurements can only be made of the entire product,
which can only provide an average value. Since x-radiography
cannot do this, the applicability of ultrasonic detection methods
was investigated during the period 2011–2014 [7].
6. Use of ultrasonic detection method
A recent review of online non-destructive inspection methods
was carried out by Kesharaju [7], and identified that ultrasonic
inspection had the potential to reduce costs, remove some of the
subjectivity and implement online automation processes. It can
also measure local variations in thickness and bulk density.
6.1. Local density variation
Kesharaju and Nagarajah [7,15–17] have recently carried out
research using an ultrasonic immersion testing methodology on
selected, defective armour components to determine local density
variations through Time-of-Flight (TOF) measurements. In this
research, the porosity dependence of ultrasonic TOF of reflected
signals was investigated to establish a correlation between the
velocity and density across the ceramic component that aids
characterisation of the RSSC material [15]. A variation in density
11590 I.G Crouch et al. / Ceramics International 41 (2015) 11581–11591
Fig. 11. X-ray of a ceramic tile (Tile 55) [7]. 1 Defect free region, 2 porosity
(density variation), 3 region containing large area of un-sintered material, 4
black spots (a mottled region containing a large number of small, isolated
pockets of unsintered material, all less than 2 mm in diameter) [7].
values between 3080 and 3160 kg/m3 was noticed with the
inspected ceramic component (see Fig. 9). The results also
indicated that the sample contained small casting-like infiltrates
and variation in grain size that occur during the high temperature
process as the liquid silicon infiltrates the green compact. The
detailed procedure followed to establish a method to quantify
density variation across ceramic components is emphasised in [15].
6.2. Manufacturing defects
In recent research [7], an ultrasonic pulse–echo contact testing
method was used to scan two different double-curved ceramic
components which had been previously subjected to X-ray
analysis. Details of the experimental procedures, along with other
ultrasonic parameters chosen, have been reported elsewhere
[15,17]. These two ceramic components were deliberately
selected because they contained a number of identifiable manu-
facturing defects: un-sintered ceramic material, free-silicon, and
porosity. The ceramic components also contained large defect-
free regions and so a database was created in order to store defect
and non-defect signal information. In the next phase, ultrasonic
signals extracted from these defects were processed further to
filter the noise and extract informative features. A learning
approach based on artificial neural networks and different signal
processing techniques was applied to classify the signals. In [16],
Kesharaju et al. used discrete wavelet transform as a pre-
processing technique to extract features from the signals.
Furthermore, to increase the classification accuracy as well as
to minimise the training time two different approaches known as
“Principal Component Analysis” [17] and “Genetic Algorithm”
[7] were investigated and the results obtained were compared at
[7]. The defects classified using the ultrasonic testing method is
illustrated in Fig. 10 below. Fig. 11 shows the corresponding X-
ray image of the same ceramic tile.
Validation of results was necessary to ensure that the signals
obtained from various defective regions corresponded to
physical defects present in the ceramic components. Similarly,
it was necessary to ensure that those signals identified as non-
defect signals actually corresponded to regions in which
defects were not present. Hence the ultrasonic results were
initially validated against X-ray results.
When comparing the ultrasonic test results against the X-rays,
defects such as large areas of un-sintered material, porosity and
defect free regions at a number of locations were detected
successfully by the ultrasonic inspection method [7]. On the
other hand, the “Black spots” (areas of large numbers of small,
isolated regions of unsintered material) shown on the X-ray
were identified as high dense regions with thickness variation
using ultrasonic testing. Hence, there is a requirement for cross
validation using another technique that confirms the results of
ultrasonic testing. Later, the results of the ultrasonic inspection
system were validated against micro-CT scan to confirm the
reliability of the experimental results and in turn the effective-
ness of the developed ultrasonic inspection system. The results
obtained were published by Kesharaju in 2014 [7].
In summary, the classification accuracy of 96% was obtained by
comparing the ultrasonic results against visual inspection and X-ray
results which were proved less sensitive in detecting defects and
measuring porosity in RSSC armour tiles in this research. Never-
theless, an overall classification accuracy of 100% was achieved
using ultrasonic inspection coupled with artificial intelligence based
signal processing methods and validated against micro-CT scan
results. It was proved, that X-ray does not lend itself to differentiating
between defects in many regions of the ceramic components tested.
This was clearly demonstrated in research conducted by Kesharaju
[7], where a porosity defect region was shown as a defect-free region
in X-ray image. In addition, the validated results demonstrate that
ultrasonic testing would be beneficial in assisting manufacturers to
check location of high porosity areas (density variation) and have the
potential for providing real time online quality control including
implementation of accept/reject criteria.
7. Conclusions
 Work carried out within the Australian defence community
over the past ten years has established comprehensive NDI
procedures for controlling the acceptable quality of RSSC
armour products. Every RSSC breastplate tile, accepted by
a range of both military and civilian clients since 2005, has
been inspected using high-resolution, digital x-radiography,
and the corresponding images stored for future reference.
This has been a powerful strategy in its own right.
 During the course of this work, a range of possible
manufacturing defects, in RSSC armour products, have been
identified and their significance established using common-
sense materials engineering principles and ballistic failure
analysis. Table 2 summarises the list of possible defects.
 I.G Crouch et al. / Ceramics International 41 (2015) 11581–11591
 Furthermore, ultrasonic detection methods, coupled with
intelligent signal processing, has been identified as being a
useful, complimentary tool for measurement of local varia-
tions in thickness and/or bulk density. It is recommended
that this method be adopted when establishing manufactur-
ing procedures in the pre-prototyping stages.
 It would appear that ultrasonic detection methods have
greater potential to be automated and the technique should
be further investigated with this objective in mind.
Acknowledgements
The authors gratefully acknowledge the work, and cooperation,
of former colleagues at both MCC and ADA with whom most of
these NDI procedures have been established. The on-going efforts
of staff at the x-radiography facility at BMT are duly acknowl-
edged for their excellent input, feedback and continued coopera-
tion in this field. The facility is currently available for on-going
research and development, as well as check-production work. The
authors also acknowledge the support given to Manasa Keshar-
aju, by the Defence Materials Technology Centre, Australia,
during the period of her Ph.D. work. The DMTC was established,
and is supported, by the Australian Government's Defence Future
Capability Technology Centre (DFCTC) initiative.
References
[1] W. Klintworth, I.G. Crouch, Cost Effective Lightweight Armour Sys-
tems, Land Warfare Conference, Adelaide, Australia, 2003.
[2] B.J. James, Practical issues in ceramic armor design, in: Proceedings of
the Ceramic Armor Materials for Design symposium, PacRim IV
International Conference on Advanced Ceramics and Glass, Maui,
Hawaii, USA, November 2001, Ceram. Trans. 134 (2002) 33–44.
[3] I.G. Crouch, Threat Defeating Mechanisms in Body Armour Systems,
Next Generation Body Armour conference, London, UK, 2009.
11591
[4] S. Leo, C. Tallon, N.A. Stone, G.V. Franks, Near-net shaping methods
for ceramic elements of (body) armor systems, J. Am. Ceram. Soc. 93
(10) (2014) 3013–3033.
[5] W.A. Gooch, An Overview of Ceramic Armor Applications, in:
Proceedings of the Ceramic Armor Materials for Design symposium,
PacRim IV International Conference on Advanced Ceramics and Glass,
Maui, Hawaii, USA, November 2001, Ceram. Trans. 134, (2002) 3–22.
[6] M.K. Aghajanian et al., A new family of Reaction Bonded Ceramics for
Armor Applications, in: Proceedings of the Ceramic Armor Materials for
Design symposium, PacRim IV International Conference on Advanced
Ceramics and Glass, Maui, Hawaii, USA, November 2001, Ceram. Trans.
134 (2002) 527–540.
[7] M. Kesharaju, Non-Contact Measurement and Characterisation of
Armour Ceramics (Ph.D. thesis), Faculty of Engineering and Industrial
Sciences,, Swinburne University, Melbourne, 2014.
[8] P.J. Hazell, Ceramic Armour: Design and Defeat Mechanisms: Miltary
Technology Series, Argos Press, Canberra, Australia, 2006.
[9] I.G. Crouch, Effects of cladding ceramics and its influence on ballistic
performance, in: Proceedings of the 28th International Symposium on
Ballistics, Atlanta, Georgia, USA, September 2014.
[10] P. Karandikar, M.K. Aghajanian, S. Wong and G. Evans, Proceedings of
the Personnel Armour Systems Symposium, Quebec, Canada, 2010,
pp. 432–441.
[11] K.M. Taylor, Cold Molded Dense Silicon Carbide Articles and Methods
of Making Same, U.S. Patent No. 3,275,722, 1965.
[12] K.M. Taylor and R.J. Palicka, Dense Carbide Composite for Armor and
Abrasives, U.S. Patent No. 3,765,300, 1973.
[13] E. Popov, Private communication (with I. G. Crouch), circa 2003.
[14] N. Haynes et al., Automated non-destructive evaluation system for hard
armour protective inserts of body armor, in: Proceedings of the 32nd
International Conference on Advanced Ceramics and Composites, Day-
tona Beach, Florida, USA, January 2008, Adv. Ceram. Armor IV. (2009)
211–218.
[15] M. Kesharaju, R. Nagarajah, Determination of density variation and
microstructure in reaction‐sintered sic ceramics using ultrasonic time-of-
flight, Int. J. Appl. Ceram. Technol. 11 (2) (2014) 303–310.
[16] M. Kesharaju, R. Nagarajah, I.G. Crouch, Ultrasonic sensor based defect
detection and characterisation of ceramics, Ultrasonics 54 (1) (2014)
312–317.
[17] M. Kesharaju and R. Nagarajah, Feature extraction for ultrasonic sensor
based defect detection in ceramic components, in: Proceedings of the 40th
Annual Review of Progress in Quantitative Nondestructive Evaluation,
Baltimore, Maryland, USA, 2014.