ASM Handbook, Volume 9: Metallography and Microstructures Copyright © 2004 ASM International®

G.F. Vander Voort, editor, p3–22 All rights reserved.

DOI: 10.31399/[Link].v09.a0003720 [Link]

Metallography: An Introduction

METALLOGRAPHY is the scientific disci- phy,” which summarizes the basic discovery by of a material or the cause of failure. This may
pline of examining and determining the consti- Sorby demonstrating the importance of speci- include:
tution and the underlying structure of (or spatial men preparation when examining metals with a ● Characterization of the macrostructural fea-
relationships between) the constituents in met- light microscope. Contrast between microstruc-
tures of a fracture surfaces to identify fracture
als, alloys and materials (sometimes called ma- tural constituents in light microscopy is very de- initiation site and changes in crack-propaga-
terialography). The examination of structure pendent on specimen preparation. Light micro- tion process
may be done over a wide range of length scales scopes also have various types of special ● Estimations of surface roughness, grinding
or magnification levels, ranging from a visual or illumination modes that can increase the infor- patterns, and honing angles
low-magnification (
20⳯) examination to mation gained from the image (see the article ● Evaluation of coating integrity and unifor-
magnifications over 1,000,000⳯ with electron “Light Microscopy” in this Volume). For ex- mity
microscopes. Metallography may also include ample, polarized-light illumination can improve ● Determination of extent and location of wear
the examination of crystal structure by tech- phase contrast, and the differential interference ● Estimation of plastic deformation associated
niques such as x-ray diffraction. However, the contrast (DIC) method can be used to identify with various mechanical processes
most familiar tool of metallography is the light topological height differences on a sample sur- ● Determination of the extent and form of cor-
microscope, with magnifications ranging from face that are smaller than 0.2 lm. rosive attack; readily distinguishable types of

50 to 1000⳯ and the ability to resolve micro- The objective of these tools is to accurately attack include pitting, uniform, crevice, and
structural features of
0.2 lm or larger. reveal material structure at the surface of a sam- erosion corrosion
The other major examination tool in metallog- ple and/or from a cross-section specimen. Ex- ● Evaluation of tendency for oxidation
raphy is the scanning electron microscope amination may be at the macroscopic, meso- ● Association of failure with welds, solders,
(SEM). Compared to the light microscope, the scopic, and/or microscopic levels. For example, and other processing operations
SEM expands the resolution range by more than cross sections cut from a component or sample This listing of macrostructural features in the
two orders of magnitude to approximately 4 nm may be macroscopically examined by light il- characterization of metals, though incomplete,
in routine instruments, with ultimate values be- lumination in order to reveal various important represents the wide variety of features that can
low 1 nm. Useful magnification covers the range macrostructural features (on the order of 1 mm be evaluated by light macroscopy.
from the stereomicroscope, the entire range of to 1 m) such as: Mesoscale structure is on the order of 1 mm
the light microscope, to much of the range of the to 100 lm. It includes microstructural features
● Flow lines in wrought products
transmission electron microscope (TEM) for at the grain level, without resolving the intrica-
● Solidification structures in cast products
possible viewing from 1,000⳯ to 100,000⳯. cies of the grain structure. For example, unifor-
● Weld characteristics, including depth of pen-
The SEM also provides a greater depth of field mity of case depth is an example of a mesoscale
etration, fusion-zone size and number of
than the light microscope, with depth of focus feature. Solidification structures at the mesoscale
passes, size of heat-affected zone, and type
ranging from 1 lm at 10,000⳯ to 2 mm at 10⳯, level include features such as cell sizes (eutectic
and density of weld imperfections
which is larger by more than two orders of mag- cell), dendrites and arms, grain type (columnar
● General size and distribution of large inclu-
nitude compared to the light microscope (Table or equiaxed), the type and concentration of
sions and stringers
1). This higher depth of field allows better dis- chemical microsegregation, and the amount of
● Fabrication imperfections, such as laps, cold
cernment of topology features during a micro- microshrinkage, porosity, and inclusions. The
welds, folds, and seams, in wrought products
scopic investigation, such as the examination of term “mesoscale” is a relatively new term, intro-
● Gas and shrinkage porosity in cast products
fracture surface during failure analysis. The duced in part to more accurately distinguish be-
● Depth and uniformity of a hardened layer in
depth of field of an SEM also may be a factor of tween different scales.
a case-hardened product
choice over light macroscopy, when very rough Microstructure is the classic term used in met-
surfaces are being examined on a macroscopic Macroscopic examination of a component sur- allography to describe features observed under a
level. For additional information on the compar- face is also essential in evaluating the condition microscope in the scale range of 1000–0.1 lm.
ative capabilities of light and electron micros-
copy, see the article “Light and Electron Mi-
croscopy” in this Volume. Table 1 Depth of field of typical light microscope objectives
However, even with the advent of electron mi-
Final magnification, Objective
croscopy, the light microscope is still the first
diameters Magnification, diameters Numerical aperture Area of field(a), lm Depth of field, lm
and most important examination device in met-
100 5.6 0.20(b) 1000 20
allography. Sometimes the contrast in a micro- 250 8.0 0.40(b) 400 3
structure is inadequate with a SEM under 500⳯, 500 21.0 0.65(b) 200 1
while it is highly visible with a basic light mi- 750 41.0 0.85(b) 135 0.4
croscope and a properly prepared sample. In- 1000 58.0 0.95(b) 100 0.1
deed, light microscopy is the historical and prac- 50.0 1.0(c) 100 0.6
1500 75.0 1.4(c) 65 0.2
tical cornerstone of metallography, as described
in the next section “The Origins of Metallogra- (a) For a final projected image 10 cm (4 in.) in diameter. (b) Dry objective. (c) Oil-immersion objective

Downloaded from [Link]

by Ben Gurion University of the Negev user
 4 / Introduction

The importance of microstructure to the prop- cause the microscopy of metals requires careful stätten structure discovered by count Alos von
erties of metals and alloys has long been recog- preparation of the surface. Unlike biological Widmanstätten, a geologist and museum curator
nized. Grain size, twins, and the size, shape, and samples, metals are opaque and thus require re- in Vienna, and his coworker Carl von Schreibers
distribution of second-phase particles are impor- flected light microscopy (where the impinging in 1808. They etched various meteorites to show
tant in determining the behavior of most metals light for viewing is reflected off the specimen the outstanding crystalline patterns in the Elbo-
and alloys. These microstructural features are surface). In contrast, biological samples are gen iron meteorite that fell in 1751. An excellent
within the fundamental resolution limits of light transparent and thus can be examined by trans- example of their work is shown in Fig. 2 (from
of 0.2 lm (or greater). Then, if necessary, the mitted light (transmission microscopy). Sorby Ref 4).
examination may move to higher levels of mag- understood the need for proper surface prepara- Widmanstätten and Schreibers etched speci-
nifications with a scanning electron microscope, tion when examining metals by reflected-light mens that could be viewed with the naked eye,
or a transmission electron microscope (TEM). microscopy. Prior to Sorby, samples were only but Sorby was the first to etch specimens and
For example, dislocations, numerous types of “distorted fractures and brutally burnished or observe the true microstructure with a micro-
second-phase particles, spinodal and ordered abraded surfaces” (Ref 2). He advised that “(the scope. Sorby first cut and polished his specimens
structures, and many aspects of martensitic final) polish must not be one which gives bright to remove all “traces of roughness.” After pol-
structures are too small for resolution by light reflection but one which may show all the irreg- ishing, he used extremely dilute nitric acid to
microscopy. Therefore, metallographic observa- ularities of the material and is as far removed as etch his specimens. He actually followed the
tion of these very fine structural features is gen- possible from a burnished surface” (Ref 3). progress of etching in order not to overetch the
erally restricted to electron microscopy. The The other piece of the metallographic puzzle specimen. The critical factor in this procedure
scale hierarchy of microstructural features is de- is the art of etching. An extremely smooth sur- was Sorby’s laborious preparation of specimens
scribed in more detail in the article “Introduction face appears nearly featureless when examined carried out by hand. The polished surfaces were
to Structures in Metals.” by reflected-light microscopy, because the light etched in dilute nitric and were undoubtedly of
reflects uniformly from the surface and appears a considerably higher standard than those of his
as a uniform contrast by human eye. Thus, tech- contemporaries, such as Wedding and Martins in
niques are needed to enhance contrast differ- Germany (Ref 5), who were also attempting to
The Origins of Metallography ences between the different phases of constitu- reveal the microstructures of steels.
ents. These methods include etching, thin-film Sorby may not have realized the exact reasons
The critical factor in the light microscopy of formation, or special illumination modes with for the success of his preparation methods, but
metals is the surface preparation of the specimen. light microscopes (see the article “Contrast En- more importantly, all of the structures reported
This is the basic insight discovered by the father hancement and Etching” in this Volume). Of by Sorby are still accepted as being correct struc-
of metallography, Henry Clifton Sorby (Fig. 1), these, chemical recipes for etching the surface tures. On 28 July 1863, Sorby recorded in his
who was the first person to examine correctly are the oldest of the various contrasting tech- diary that he had “discovered” the structure of
polished and chemically etched metal samples niques. Etching even precedes Sorby by at least an iron. It was not until 1886 and 1887, however,
under the microscope in 1863 (Ref 1). This ap- four centuries, as in the case of macroetching that his results were recorded in a journal with a
plication of microscopy is more than two cen- techniques to reveal the damask patterns of wide readership (Ref 1, 3). By careful observa-
turies later than the biological microscope, be- swords and various pieces of armor. Macroetch- tion he identified major microstructural constit-
ing was also used to reveal the structure of pol- uents of ferrous materials (the constituents now
ished meteorites, such as the famous Widman- known as graphite, cementite, pearlite,* austen-
ite, and the phosphide eutectic). He recognized
that iron was composed of a number of crystal
grains, and he also realized that iron underwent
an allotropic change on heating. As noted by
Samuels (Ref 6), these are awesome achieve-
ments considering that he started from scratch
and that they were achieved after such a short
period of investigation.
From this beginning, the importance of spec-
imen preparation remains central today. Many
deficiencies arise when the preparation methods
are neglected. False structures (or artifacts) can
arise from the preparation in many ways. In par-
ticular, Jose Ramon Vilella (Fig. 3) was the first
to realize that artifacts were sometimes being ob-
served due to the presence of a layer of “distorted
or disturbed” metal formed during the early
stages of surface preparation and not during pol-
ishing itself (Ref 7). He demonstrated that the
true microstructure was seen only when the dis-

*At the end of the 19th century, very fine pearlite unresolved
Fig. 2 Macrograph of the Elbogen iron meteorite pre- in the light microscopes was referred to as “sorbite” in honor
pared in 1808 by Widmanstätten and Schreibers of Sorby. However, because it is not a new constituent, the
using heavy etching in nitric acid. After rinsing in water term “sorbite” did not survive. The term “pearlite” survives to
Fig. 1 Henry Clifton Sorby (1826–1908), geologist, pe- and drying, printer’s ink was rolled on the etched surface, this day and is actually connected to Sorby, because he de-
trographer, mineralogist, and founder of metal- and the sample was pressed onto a piece of paper. Source: scribed the “pearly constituent,” i.e., pearlite, as having a
lography. Source: Ref 6 Ref 4 “mother of pearl” appearance.

Downloaded from [Link]

by Ben Gurion University of the Negev user
 Metallography: An Introduction / 5

turbed layer was removed, and he devised a in the next Section “Metallurgy and Microstruc- revealed by applying a dye penetrant to the pol-
method (alternate etching and polishing) of do- ture.” Interpretation of micrographs also requires ished specimen. The dye was drawn into the
ing this (Fig. 4) (Ref 7). an understanding of how specimen preparation cracks by capillary action, and the surface was
Vilella’s seminal work established the need for and microscopic techniques affect the appear- then wiped clean. The specimen was then placed
preparation procedures beyond just the produc- ance of particular phases in a given material. under a light that caused the dye to fluoresce,
tion of reflecting surfaces. Successful metallog- Thus, the cataloging of micrographs (in print and the cracks became readily observable. Dye-
raphy imposes the following requirements on the and/or electronic form) can be useful when com- penetrant techniques are excellent for examina-
final preparation of the specimen surface (Ref 6): paring the effects of material variations and tion of cracklike macrostructural imperfections
changes in specimen preparation. in metals. However, grains and other microstruc-
● Surface layers that might obscure structural
tural features are visible only after etching,
features must not be present.
which frequently obscures the presence of the
● False structures that might be detected during
a subsequent examination must not have been Macroanalysis (Adapted from Ref 8) cracks. Therefore, different metallographic tech-
niques are necessary to reveal various macro-
introduced.
Macrostructural characterization of metals structural elements.
● All desired fields of view must be coplanar
within the depth-of-field limits of the system and alloys is the detailed evaluation of large-
to be employed for examination. scale inhomogeneities in composition, morphol- Macroscopy of Sections
● The surface must be adequately free from ogy, and/or density. These inhomogeneities may
stains and other accidental blemishes. develop during such procedures as casting, ex- Preparation of a metallographic section for ex-
trusion, forging, rolling, and welding or during amination requires careful selection of the area
With these basic objectives in mind, then the service. Macroscale examination of surfaces is to be characterized (see the article “Metallo-
next question is determining the most effective also essential in the failure analysis of fractured, graphic Sectioning and Specimen Extraction” in
mechanical, chemical, and/or physical methods corroded, and/or worn parts. Microscopic eval- this Volume). This area must be chosen to rep-
of specimen preparation for the appropriate mi- uation clearly is a significant step in any failure resent the unique features of a specific zone of
croscopic tool. These methods are described in examination, but it should not replace character- interest or the general features of a part or com-
more detail in the Section “Metallographic Tech- ization by macroscopy. These two types of met- ponent selected for process characterization or
niques” of this Volume. allography are complementary, but examination quality assurance. The selected region of the
After a micrograph from a properly prepared during failure analysis should always begin at specimen must then be removed from the com-
specimen is obtained and recorded, the next low magnification and work upward. A frequent ponent using techniques that do not damage or
challenge is to interpret, understand, and use the mistake in failure analysis is to neglect exami- distort the features of interest. The section of in-
information contained on the recorded image. In- nation of the broken pieces at low magnifica- terest is then prepared metallographically, and
terpretation of microstructural features requires tions. Too frequently, the component is sectioned the prepared section is characterized using mac-
an understanding of crystal structure, kinetics, immediately. roscopic examination.
and the metallurgical mechanisms of solidifica- Examination techniques other than metallog- Macroscopic examination generally does not
tion, deformation, and phase transformations. raphy may also be more effective during mac- require the extreme surface smoothness needed
These topics, as they relate to structure, are in- roscale examination. For example, Fig. 5 shows for microscopic examinations. Such surface
troduced in more detail in the series of articles spider cracks in the center of a copper specimen. preparation techniques as etching are frequently
This specimen was sectioned, ground, and pol- prolonged such that surface features are greatly
ished, but not etched. Chemical etching and sub- enhanced; therefore, quantitative measurements
sequent evaluation of the macrostructure may should not be conducted on macroetched sam-
fail to reveal this type of structural imperfection ples. Heavy etching accentuates any microstruc-
(Fig. 5b). The cracks shown in Fig. 5(a) were tural inhomogeneity (Fig. 6). The flow lines

Fig. 3 Jose Ramon Vilella (1897–1971), distinguished Fig. 4 An example used by Vilella to illustrate the effect of disturbed metal on the appearance of pearlite. (a) Polished
metallographer who understood the need to surface covered by a layer of disturbed metal; structures such as this were called sorbite or troostite-sorbite by
faithfully prepare representative surfaces in metallographic some early investigators. (b) Same field after removing the layer of disturbed metal by alternate polishing and etching;
examinations. Source: Ref 6 true structure of lamellar pearlite. Etched in picral reagent. 1000⳯. Source: Ref 7

Downloaded from [Link]

by Ben Gurion University of the Negev user
 6 / Introduction

show the direction of metal flow during process- Features often associated with the fracture sur- and/or low-temperature service (deformation
ing and frequently represent paths for easy frac- face at the macroscale and microscale are shown twins), and so forth. This also is very necessary
ture. Figure 7 shows the use of similar macro- in Tables 2 and 3. Examination of the informa- to the failure investigation.
scopic techniques to illustrate the depth of case tion in these tables shows that the fracture fea- Macroscopic features typically help identify
hardening in a tool steel. Figure 8 is a weld mac- tures provide information about (Ref 9): the fracture-initiation site and crack-propagation
rograph that shows the different etching char- direction. The orientation of the fracture surface,
● The crack-initiation site and crack-propaga-
acteristics of the fusion zone and heat-affected the location of crack-initiation site(s), and the
tion direction
zone (HAZ) of a weld. The 2% nital etchant used
● The mechanism of cracking and the path of
to reveal the weld macrostructure is much less
fracture
aggressive than the 50% hydrochloric acid etch-
● The load conditions (monotonic or cyclic)
ants used on the specimens shown in Fig. 6 and
● The environment
7 and reveals finer structural detail but requires
● Geometric constraints that influenced crack
a polished specimen.
initiation and/or crack propagation
In castings, macroscopy is used to establish
● Fabrication imperfections that influenced
the outer chill zone depth, shape, and size of the
crack initiation and/or crack propagation
columnar or dendritic grains perpendicular to the
mold wall, and size of the central equiaxed zone It should also be clear that not all features cre-
(Fig. 9). For example, Fig. 10 shows the ma- ated by a given cause for failure are necessarily
crostructure of a small, relatively pure aluminum present on a given fracture surface. For example,
ingot exhibiting typical cast grain structure. To beach marks (at low magnification) and stria-
obtain the macrograph, the aluminum ingot was tions (at higher magnification) are well-known
sectioned, then ground and polished to produce features of fatigue cracks, but are not always
a flat reflective surface. The polished section was present or visible. In addition, not all fracture
then etched by immersion in a solution that at- mechanisms have unique appearances. For ex-
tacked the various grain orientations at different ample, intergranular fracture can be caused by a
rates. The structural elements visible in this mac- number of mechanisms.
rograph are grains. The small grains near the bot- It is also important to understand that the frac-
tom of the ingot appear relatively equiaxed. This ture surface only provides evidence of the crack-
region of small equiaxed grains is the chill zone. propagation process; it does not reveal evidence
Macroscopy of cast structures is also used to re- of events prior to nucleation and growth. Ex-
veal imperfections such as shrinkage, gas, po- amination beyond the fracture surface also pro-
rosity, and center cracks. vides information. For example, visual inspec-
tion of a fractured component may indicate
events prior to fracture initiation, such as a shape
change indicating prior deformation. Metallo-
Macroscopy of Fracture graphic examination of material removed far
Surfaces (Adapted from Ref 9) from the fracture surface also can provide infor-
mation regarding the penultimate microstruc-
Both the macroscale and microscale appear- ture, including the presence of cold work (slip,
ances of fracture-surface features can tell a story bent annealing twins, deformation bands, and/or
of how and sometimes why fracture occurred. grain shape change), evidence of rapid loading

Fig. 6 Flow lines in a forged 4140 steel hook. Specimen

was etched using 50% HCl. 0.5⳯. Source: Ref 8

Fig. 7 Case-hardened layer in W1 tool steel. Specimens

were austenitized at 800 C (1475 F), brine
Fig. 5 Macrostructure of a continuous-cast copper ingot. (a) Spider cracks revealed using dye-penetrant inspection. quenched, and tempered 2 h at 150 C (300 F). Black rings
Transverse section at top; longitudinal section at bottom. (b) Same ingot, etched using Waterbury’s reagent. are hardened zones. Etched using 50% hot HCl. Approxi-
Cracks are not revealed. Both approximately 0.5⳯. Source: Ref 8 mately 0.5⳯. Source: Ref 8

Downloaded from [Link]

by Ben Gurion University of the Negev user
 Metallography: An Introduction / 7

crack-propagation direction should correlate the failed component is in multiple pieces, and fracture. That might not be the case for a lami-
with the internal state of stress created by the chevrons are visible on the fracture surface, anal- nated structure loaded in bending.)
external loads and component geometry. When ysis of crack branching (crack bifurcation) (Fig. Likewise, the profile of a fracture surface rela-
11) (Ref 10) can be used to locate the crack- tive to loading direction can indicate the mode
initiation site. Fracture initiates in the region of fracture by elastic (plane-strain) conditions or
where local stress (as determined by the external elastic-plastic (plane-stress) conditions. Plane-
loading conditions, part geometry, and/or mac- strain (or mode I) fracture is characterized by a
roscopic and microscopic regions of stress con- flat surface perpendicular to the applied load.
centration) exceeds the local strength of the ma- Plane-stress (mode II) fracture occurs when
terial. Thus, variations in material strength and shear strain becomes the operative mode of de-
microscale discontinuities (such as an inclusion formation and fracture (as maximum stresses oc-
or forging seam) must be considered in conjunc- cur along the shear plane from the basic princi-
tion with variations in localized stress that is de- ples of continuum mechanics). In plane-stress
termined by applied loads and macroscopic cracking, the fracture profile is characterized by
stress concentrations (such as a geometric notch shear lips, which are at about a 45 oblique angle
or other change in cross section). to the maximum stress direction (although this
The fracture surface orientation relative to the angle may vary depending on material condition
component geometry may also exclude some and loading condition). In general, these varia-
loading conditions (axial, bending, torsion, tions in fracture profiles are related to fracture
monotonic versus cyclic) as causative factors. toughness, which depends on section thickness
For example, crack initiation is not expected (B) and the size (a) of a preexisting discontinuity
along the centerline of a component loaded in such as a notch. This is shown in Fig. 12. Crack-
bending or torsion, even if a significant material tip radius also influences fracture behavior.
imperfection is present at that location because Surface roughness and optical reflectivity also
no normal stress acts at the centerline. (There is provide qualitative clues to events associated
a shear stress at this location in bending, but in with crack propagation. For example, a dull/
a homogeneous material, it is too small to initiate matte surface indicates microscale ductile frac-
Fig. 8 Section through an arc butt weld joining two 13
mm (0.5 in.) thick ASTM A517, grade J, steel
plates. Etched using 2% nital. 4⳯. Source: Ref 8
Table 2 Macroscale fractographic features
Mark/Indication Implication
Visible distortion Plastic deformation exceeded yield strength and may indicate
instability (necking, buckling) or post-failure damage
Visible nicks and gouges Possible crack initiation site
Fracture surface orientation relative to component • Helps separate loading modes I, II, III
geometry and loading conditions • Identifies macroscale ductile and brittle fracture.
Both flat fracture and shear lips present on fracture • Crack propagation direction parallel to shear lips
surface • Mixed-mode fracture (incomplete constraint)
Tightly closed crack on surface • Possible cyclic loading
• Possible processing imperfection, e.g., from shot peening, quench
cracks
Radial marks and chevrons (V-shape) • Point toward crack initiation site
• Show crack propagation direction
Fig. 9 Sketch of grains in a typical cast ingot Crack arrest lines (monotonic loading) (U-shape) • Lines point in direction of crack propagation
• Indicate incomplete constraint
Crack arrest lines (cyclic loading) (beach marks, • Indicates cyclic loading
conchoidal marks) • Propagation from center of radius of curvature
• Curvature may reverse on cylindrical sections as crack
propagates
Ratchet marks • More likely in cyclic loading
• Indicates initiation site(s)
Adjacent surface and or fracture surface discoloration • May indicate corrosive environment
• May indicate elevated temperature
Oxidized fingernail on fracture surface Possible crack initiation site
Fracture surface reflectivity • Matte: ductile fracture or cyclic loading
• Shiny: cleavage likely
• Faceted (“bumpy”) and shiny; intergranular fracture in large
grain size
Fracture surface roughness • Increase in surface roughness in direction of crack growth (may
be affected in bending by compressive stressed region when
crack moves into this region)
• Smooth region plus rough region in direction of growth—cyclic
loading
• Rough matte fractures are ductile
• May indicate transition from fatigue crack growth to overload
Rubbing (general) • May indicate vibration
• May show final direction of separation
• Swirl pattern indicates torsion
Rubbing (localized) • May indicate crack closure in cyclic loading
• May obliterate beach marks
Fig. 10 Macrostructure of as-cast aluminum ingot. Deformed draw marks, rolling scratches If twisted, indicates torsion loading
Transverse section shows outer chill zone and Machining marks (normal to axis of component) Not distorted in torsion loading
columnar grains that have grown perpendicularly to the Variable roughness of fracture edge In brittle bending, rough side is tension side
mold faces. Etched using Tucker’s reagent. 1.5⳯. Source:
Source: Ref 9
Ref 8

Downloaded from [Link]

by Ben Gurion University of the Negev user
 8 / Introduction

ture, while a shiny, highly reflective surface in- fracture of a pearlitic steel or annealed copper. defined in castings. The “V” of a chevron points
dicates brittle cracking by cleavage or Also, fracture surface roughness increases as a back to the initiation site, and a sequence of “V”s
intergranular fracture. In addition, when inter- crack propagates so the roughest area on the frac- across the fracture surface indicates the crack-
granular fracture occurs in coarse-grained ma- ture surface is usually the last to fail. Fracture propagation direction. The appearance of chev-
terials, individual equiaxed grains have a dis- surface roughness and the likelihood of crack bi- rons or radial marks near the crack origin de-
tinctive rock-candy appearance that may be furcation also increase with magnitude of the ap- pends in part on whether the crack-growth
visible with a hand lens. In terms of document- plied load and depend on the toughness of the velocity at the surface is greater or less than that
ing surface conditions, one major problem with material. Brittle failures often contain multiple below the surface. If crack-growth velocity is at
optical (light) macroscopic or microscopic ex- cracks and separated pieces, while ductile over- a maximum at the surface, radial marks have a
amination of fracture surfaces is its inability to load failures often progress as single cracks, fan-shaped appearance (Fig. 13). If crack-growth
obtain favorable focus over the entire surface if without many separated pieces or substantial rate is greatest below the surface, the result is
the magnification exceeds 5 to 10⳯. Therefore, crack branching at the fracture location. chevron patterns (Fig. 14).
SEM also has become a standard metallographic Under the right conditions, fracture surfaces In rectangular sections, specimen dimensions
tool in failure analysis. may also have radial marks and chevrons, which can affect the appearance of radial markings and
Surface roughness provides clues as to are macroscopic surface features that indicate the chevron patterns. For example, the macroscale
whether the material is high strength (smoother) region of crack initiation and propagation direc- fracture appearances of unnotched sections are
or low strength (rougher) and whether fracture tion. They are common and dominant macro- shown in Fig. 15 for sections with various width-
occurred as a result of cyclic loading. The sur- scopic features of the fracture of wrought metal-
faces from fatigue crack growth are typically lic materials, but are often absent or poorly
smoother than monotonic overload fracture ar-
eas. The monotonic overload fracture of a high-
strength quenched-and-tempered steel is signifi-
cantly smoother overall than is the overload

Fig. 13 Radial marks typical of crack propagation that

is fastest at the surface (if propagation is unin-
fluenced by the configuration of part or specimen)

Fig. 14 Chevron patterns typical when crack propa-

gation is fastest below the surface. It is also ob-
Fig. 11 Component that has fractured in multiple served in fracture of parts having a thickness much smaller Fig. 15 Typical fracture appearances for unnotched
pieces. If chevrons are visible on the fracture than the length or width (see middle illustration in Fig. 15). prismatic tension-test sections. Source: Ref 9
surface, the sequence of crack formation can be used to
obtain the crack formation of sequence and the location of
the initiation site. Source: Ref 10
Table 3 Microscale fractography features
Mark/Indication Implication
Dimpled fracture surface Ductile overload fracture at this location
Faceted fracture surface • Brittle cleavage fracture
• Possible SCC fracture
• May be low DK fatigue
Intergranular with smooth grain boundaries • Likely either improper thermal processing or environmental
assisted fracture (high temperature, corrosive environment)
• Less common is low DK fatigue
Intergranular with dimpled grain boundaries • “Decohesive” rupture—fracture at high fraction of melting point
• Possible improper processing creating denuded zone adjacent to
grain boundary
River pattern or fan pattern Cleavage fracture; crack runs “down” river; fan rays point to
initiation site within a grain.
Tongues Twinning deformation during rapid crack propagation
Flutes on transgranular fracture surface • Indicates corrsive environment and ductile fracture
• Crack propagates parallel to flutes.
Striated or ridged fracture • Cyclic loading fatigue striations; Constant spacing, constant
stress amplitude; variable spacing, variable stress amplitude or
block loading
• Striated surface caused by second phases in microstructure.
Fig. 12 Variation in fracture toughness and macroscale Grooves or flutes • SCC
features of fracture surfaces for an inherently • Transgranular fracture
ductile material. As section thickness (B) or preexisting Artifacts (mud cracks) Dried liquid on surface. May indicate incomplete cleaning of
crack length (a) increases, plane-strain conditions develop surface. If in the as-received condition, may indicate fluids from
first along the centerline and result in a flat fracture surface. service and may indicate SCC conditions. Material should be
With further increases in section thickness or crack size, analyzed.
the flat region spreads to the outside of the specimen, de- Artifacts (tire tracks) • Common in cyclic loading
creasing the widths of the shear lips. When the minimum • Due to entrapped particulate matter
value of plane-strain toughness (KIc) is reached, the shear
Source: Ref 9
lips have very small width. Source: Ref 9

Downloaded from [Link]

by Ben Gurion University of the Negev user
 Metallography: An Introduction / 9

to-thickness (w/t) ratios. The w/t ratio influences or rosette, is perpendicular to the crack front and, specimen. Ductile fractures often progress as
the ability of the sample to maintain a unidirec- as such, may be considered to be the round-sam- single cracks, without many separated pieces or
tional state of stress during tension. In a thick ple equivalent of the radial markings or chevron substantial crack branching at the fracture loca-
section (top), strain in the width direction is con- patterns that appear on sheet or plate samples, as tion. The region of a crack-initiation typically
strained and thus tends to a condition of plane- previously described. The radial marks, or radial has a dull fibrous appearance that is indicative
strain (mode I) fracture. In this case, a large por- shear marks, are visually distinct from the fi- of cracking by MVC. The crack profiles adjacent
tion of the fracture surface comprises radial brous region, as shown in Fig. 17 for an un- to the fracture are consistent with tearing. The
markings or chevron patterns indicative of rapid, notched SAE 4150 steel specimens with differ- fracture surface may have radial markings, chev-
unstable cracking. At higher w/t ratios, the radial ent strengths. Figure 17(a) shows a clear rons, and/or shear lips depending on the speci-
zone is suppressed in favor of a larger shear-lip boundary between the fibrous central region and men geometry and material condition, as previ-
zone. In very thin sections (bottom), plane-stress the large ridge pattern of the radial marks. Figure ously noted.
conditions apply, and the fracture surface is com- 17(b) shows shallower radial marks and a An example of mixed-mode (mode I and II)
posed almost entirely of a shear lip outside the slightly larger fibrous zone from a heat treatment fracture is the classic cup-and-cone appearance
fibrous zone of crack initiation. Figure 16 shows that results in more ductility. Figure 17(c) shows from ductile fractures of unnotched cylindrical
radial marks and chevrons when fracture initiates very weak or shallow radial marks that develop tension-test specimen (Fig. 18a). In this case, the
from surface notches. further from the center. fracture originates near the specimen center,
If conditions are right, the radial patterns as- Radial marks on the fracture surface of an un- where hydrostatic stresses develop during the
sociated with rapid or unstable crack propagation notched cylindrical tension-test specimen (Fig. onset of necking and where microvoids develop
can also occur in a cylindrical section. This ra- 18) point to the center, which has a fibrous ap- and grow. Multiple cracks join and spread out-
dial pattern, sometimes called a radial shear, star, pearance that is associated with ductile crack ini- ward along the plane normal to loading axis, as
tiation and growth by microvoid coalescence. representative of mode I (plane-strain) crack
However, if the specimen is notched (Fig. 18b), propagation. When cracks reach a region near
then crack initiation may begin at several loca- the outer surface, the mode of fracture changes
tions along the circumference near the root of to mode II (plane-stress) condition, where shear
the notch, where stress concentration occurs. The strain becomes the operative mode of deforma-
region of crack initiation may still have a fibrous tion. Thus, even though the overall applied stress
appearance indicative of microvoid coalescence is still a tensile load, deformation makes a tran-
(MVC) near the root of the notch, but the region sition to the shear plane in the outer regions of
of final, fast fracture is in the center and roughly the specimen and thus results in the 45 shear
perpendicular to the applied load. Thus, even lips that are indicative of a mode II fracture. Al-
though the radial markings may appear to point ternatively, the fracture mode may be entirely
to the center, the surface conditions indicate that plane strain when a sufficiently large crack or
the central region is the area of final fracture, not notch is introduced (Fig. 18b).
crack initiation. In effect, the notch size is suf- Macroscopic Appearances of Brittle Frac-
ficient to cause plane-strain fracture, as evi- tures. Brittle overload failures, in contrast to
denced by the lack of shear lips. ductile overload failures, are characterized by lit-
Macroscopic Appearance of Ductile Frac- tle or no macroscopic plastic deformation. Brittle
Fig. 16 Typical fracture appearances for edge- and
tures. As noted in Table 2, ductile fractures are fracture initiates and propagates more readily
side-notched rectangular tension-test sections.
Note the shear lips when the fracture approaches the edge typically characterized by evidence of plastic de- than ductile fracture or so-called “subcritical”
of the specimen. Source: Ref 9 formation, such as necking of a tension-test crack-propagation processes such as fatigue or

Ultimate tensile Charpy V-notch

Yield strength strength impact energy
Fibrous zone as
Specimen Hardness, HV MPa ksi MPa ksi Reduction of area, % J ft • lbf percentage of total area
(a) 285 0.73 0.106 0.83 0.120 66 163 120
25
(b) 258 0.65 0.094 0.79 0.115 67 174 128
31
(c) 301 0.81 0.117 0.97 0.141 49 27 20
44

Fig. 17 Radial marks on tensile test specimen of Society of Automotive Engineers (SAE) 4150 steel isothermally transformed to bainite, quenched to room temperature, and then
tempered. (a) Lower bainite, isothermally transformed at 300 C (570 F) for 1 h, tempered at 600 C (1110 F) for 48 h. (b) Lower bainite, isothermally transformed at 375
C (705 F) for 1 h, tempered at 600 C (1110 F) for 48 h. (c) Upper bainite, isothermally transformed at 450 C (840 F) for 24 h, as-quenched. Source: David Johnson, Master’s thesis,
University of Tennessee

Downloaded from [Link]

by Ben Gurion University of the Negev user
 10 / Introduction

stress-corrosion cracking. Because brittle frac- new surfaces, that is, cracks. For this reason, marks that are initially fairly concentric to the
tures are characterized by relatively rapid crack brittle failures often contain multiple cracks and origin.
growth, the cracking process is sometimes re- separated pieces, which are less common in duc- Macroscopically, fatigue fracture ordinarily
ferred to as being “unstable” or “critical” be- tile overload failures. Brittle fracture mecha- has a brittle appearance and lacks the gross plas-
cause the crack propagation leads quickly to final nisms may exhibit chevron or herringbone pat- tic deformation (e.g., necking) characteristic of
fracture. terns that indicate the fracture origin and ductile tensile overload fracture. In contrast with
The macroscopic behavior is essentially elas- direction of rapid fracture. Chevrons occur ductile overload fracture, which generally has
tic up to the point of failure. The energy of the mainly in structural steels and rail (web) steel or more-or-less shear lip (slant 45 fracture) along
failure is principally absorbed by the creation of relatively ductile low-strength alloys. Chevrons free surfaces, propagating fatigue fractures typ-
are dependent on strength, ductility, and section ically intersect free surfaces at right angles (Fig.
thickness, and are not normally seen in high 20). This provides a tool for helping to identify
strength alloys. fatigue locations. In common with other pro-
Herringbone patterns are unique microscopic gressive fracture modes, such as stress-corrosion
features of brittle fractures (Ref 9). Ductile cracking, field fatigue fractures are frequently
cracking, which occurs by microvoid coales- decorated by more-or-less curved marks that de-
cence, does not result in a herringbone pattern. lineate the position of the crack front at a partic-
On a microscopic scale, the features and mech- ular point in time. These marks are commonly
anisms of fracture may have components of duc- called beach marks and are also known as clam-
tile or brittle crack propagation, but the macro- shell marks or arrest marks.
scopic process of fracture is characterized by Beach marks are produced by a change in
little or no work being expended from defor- crack-growth conditions, such as a change in en-
mation. vironment or stress level or a pause in stress cy-
Macroscopic Appearances of Fatigue Frac- cling (interruption in service). Thus, beach
tures (Adapted from Ref 11). Examination of marks are not always present on the surface of a
a fatigue fracture usually begins with unaided fatigue fracture. For example, beach marks are
visual observation, often followed by viewing not found in laboratory tests conducted under
with a hand lens or stereomicroscope. Macro- uniform loading and environmental conditions
scopic examination of fracture surfaces can be (e.g., Fig. 20). Moreover, the presence of beach
performed on-site (when the broken part is ac- marks also is not conclusive evidence of fatigue
cessible), requires little or no preparation of the fracture. Beach marks may also appear when
specimen, and uses minimal and relatively sim- fracture is from stress-corrosion cracking Fig. 21
ple equipment. It does not destroy the specimen (Ref 12).
or alter fracture surfaces. Macroscopic exami-
nation is particularly useful in correlating frac-
ture surface characteristics with part size and Microscopic Examination
shape and with loading conditions.
Fatigue origins are frequently located most The importance of microstructure to the prop-
Fig. 18 Fracture surface regions in cylindrical tension-
readily by viewing the fracture surface at low erties of metals and alloys has long been recog-
test specimens. (a) Surface from cone portion
of fractured unnotched tensile specimen. (b) Surface of magnifications (up to 30 to 50⳯). For example, nized. Grain size, twins, and the size, shape, and
fractured notched specimen. Unlike the fracture surface for Fig. 19 shows a fracture of a steel housing tube. distribution of second-phase particles are impor-
an unnotched specimen, the fracture surface for the The initiation region is observable in the mac- tant in determining the behavior of most struc-
notched specimen (b) does not have shear lips, because
the fracture initiates near the root of the notch (and com-
rograph, as shown by the arrow. The position of tural metals. Therefore, characterization of mi-
pletely around the specimens in this idealized case without the crack front at various times during the failure crostructures by light microscopy is essential.
additional stress raisers). Source: Ref 9 process is also visible as the so-called beach Process-control parameters are established to

Fig. 19 Fractographs of a typical fatigue crack in a clamp. (a) The fatigue crack origin is marked by the arrow. The crack propagated to the right by continuous fatigue cracking
(light) region, then continued alternately by rapid tearing and slow fatigue cracking. 2⳯. (b) Higher-magnification view of the region near the arrow in (a). 10⳯. Source:
Ref 8

Downloaded from [Link]

by Ben Gurion University of the Negev user
 Metallography: An Introduction / 11

provide specific grain sizes. The number, size, Microscopy is also essential in the analysis of are two types of replicas: surface replicas and
and distribution of second-phase particles, such failures due to fracture, wear, and/or corrosion. extraction replicas. Surface replicas provide an
as inclusions, are frequently specified, and quan- These topics and the use of light and electron image of the surface topography of a specimen,
titative metallographic procedures have been de- microscopy in failure analysis are discussed in while extraction replicas lift particles from the
veloped to describe microstructure. more detail in Failure Analysis and Prevention, specimen. The application of replica metallog-
This section on microanalysis focuses mainly Volume 11 of ASM Handbook. However, the sec- raphy is discussed in more detail in the article
on the method of light microscopy with some tion “Microfractography” in this article briefly “Field Metallography Techniques” in this Vol-
discussion of SEM in fractography. As previ- compares the application of light and electron ume.
ously noted, the upper limit of useful magnifi- microscopy in fractographic analysis. Another
cation in a light microscope is approximately important technique in microanalysis is replica
1500⳯, and the fundamental limitations of light metallography, where specimens are replicas Microstructure
optic systems limit resolution to features that are taken in situ from components in the field. There
Optical (light) characterization of the micro-

0.2 lm or larger. Light microscopy, then, is
structures of metals and alloys involves the iden-
used primarily to examine grain structures and
tification and measurement of phases, precipi-
the morphology of large second-phase particles.
However, many other microstructural features tates, and constituents, and the determination of
that are too small to be observed using light mi- the size and shape of the grains, the extent of
croscopy also can influence the properties of twinning, and some of the characteristics of
metals and alloys. Dislocations, numerous types grain boundaries and other observable defects.
of second-phase particles, spinodal and ordered Solidification, solid-state transformation, defor-
structures, and many aspects of martensitic mation, and annealing microstructures are the
structures can be categorized as too small for four basic types in metals and alloys. Each of
light microscopy. These features require exami- these has distinct characteristics, as described
nation by electron microscopy, which are dis- below. Anisotropy of grain orientation is also
cussed elsewhere (see the articles “Scanning important when characterizing the microstruc-
Electron Microscopy” and “Transmission Elec- ture of a material.
tron Microscopy” in this Volume). Anisotropy. Microstructural features exist in
three dimensions, while metallographic obser-
vation typically represents only two dimensions.
Therefore, effective microscopy frequently re-
quires microstructural observations in two or
more directions. For example, Fig. 22 and 23
illustrate the value of viewing the microstructure
in several directions. Figure 22 shows an an-
nealed microstructure exhibiting similar grain
shapes in all three views. Grain size is charac-
terized by placing a line of known length (or
preferably a circle of known circumference) on
the magnified image of the microstructure and
counting the number of intersections between
the line and grain boundaries in the microstruc-
Fig. 20 Aluminum alloy fracture mechanics test spec- ture. The number of grain boundary intersec-
imen, 6.3 mm (0.25 in.) thick. Fatigue crack at Fig. 21 Beach marks on a 4340 steel part caused by tions, P, can be converted to a measure of grain
left of arrows is flat and perpendicular to side surfaces (note stress-corrosion cracking. Tensile strength of size, l, using:
absence of beach marks in this laboratory fatigue fracture). the steel was approximately 1780 to 1900 MPa (260 to 280
Overload fracture to right of arrows has 45 shear lips ex- ksi). The beach marks are a result of differences in the rate
L
tending upward at the top side of the sample and down- of penetration of corrosion on the surface. They are in no l=
ward at the bottom side. Source: Ref 11 way related to fatigue marks. 4⳯ PM (Eq 1)

Fig. 22 Copper alloy 26000 (cartridge brass, 70%) sheet, hot rolled to a thickness of 10 mm (0.4 in.), annealed, cold rolled to a thickness of 6 mm (0.230 in.), and annealed to a
grain size of 0.120 mm (0.005 in.). At this reduction, grains are basically equiaxed. Compare with Fig. 23. Diagram in lower left of each micrograph indicates orientation
of the view relative to the rolling plane of the sheet. Etched using NH4OH plus H2O2. 75⳯. Source: Ref 8

Downloaded from [Link]

by Ben Gurion University of the Negev user
 12 / Introduction

where M is the magnification of the image ob- compositional changes (termed “coring”), the engineering reliability. Quality-assurance pro-
served, L is the length of line on the image, and rate of etching at interdendritic regions differs grams frequently require controlling imperfec-
grain size (l) is the mean lineal intercept length from that at dendrite arms. If the alloying ele- tions to regulate their type, number, size, and
(per ASTM E112). The microstructure of the an- ment or impurity content is high, interdendritic shape in a particular manner. For example, a
nealed alloy (Fig. 22) is isotropic, while the regions may develop a two-phase structure. Be- component having a stringer distribution such as
grains are elongated in the rolling direction and cause dendrite arm spacing tends to decrease that shown in Fig. 25(c) would have better duc-
flattened in the transverse directions when alloy with increasing cooling rates, the properties of tility if specimens or components were tested
is in the cold-rolled condition. This anisotropic as-cast metals depend on the solidification rates. with the major stresses parallel to the stringer
grain structure also renders anisotropic mechan- Most metals shrink during solidification. than if specimens were oriented with the major
ical properties. Physical properties may also be Therefore, the liquid trapped between dendrite stresses perpendicular to the stringer.
anisotropic, especially in single-phase alloys due arms during solidification is frequently insuffi- Transformation structures often consist of
to texture. Thus, anisotropic materials may re- cient to fill the space between the arms when two phases. In such structures, the major phase
quire selection, preparation, and viewing of solidification is complete. This inability to fill is typically termed the matrix, or base structure,
specimens from different orientations. This is the remaining space leads to shrinkage voids, and the minor phase is termed the second phase.
discussed further in the article “Metallographic which can be observed microscopically. Voids The size, shape, and distribution of second-phase
Sectioning and Specimen Extraction” in this are generally easier to observe on as-polished particles are important in determining the prop-
Volume. Modern techniques also include meth- specimens than on polished and etched ones. erties of metals and alloys. Characterization of
ods of three-dimensional representation (see the Figure 25(a) shows a typical example of shrink- second-phase morphology can sometimes be ac-
article “Three-Dimensional Microscopy” in this age voids. complished using optical metallography. How-
Volume). Discontinuities. Various materials disconti- ever, the second phase is sometimes so small that
Solidification Structures. The most com- nuities, such as inclusions and stringers (Fig. 25b the resolution necessary to characterize the phase
monly observed solidification structure is den- and 25c) can also be observed microscopically morphology exceeds the limits of the light mi-
dritic (Fig. 24). A dendritic structure usually ex- in as-polished specimens. Such imperfections as croscope. In these cases, SEM may be used, or
hibits compositional variations, with the those shown in Fig. 25 can serve as failure-ini- transmission electron microscopy may be
primary, secondary, and tertiary dendrite arms tiation sites in metals and alloys; therefore, char- needed. Age-hardenable or precipitation-hard-
containing less alloying or impurity elements acterization of their size, shape, and distribution ened metals and alloys generally must be char-
than interdendritic regions. Because of such is necessary to establish material properties and acterized using electron microscopy.

Fig. 23 Same alloy and processing as in Fig. 22, but reduced 50% by cold rolling from 6 mm (0.239 in.) to 3 mm (0.120 in). Grains are elongated in the rolling direction. Diagrams
indicate same orientation of view as in Fig. 22. Etched using NH4OH plus H2O2. 75⳯. Source: Ref 8

Fig. 24 Dendritic solidification structure in a Ni-5Ce (at.%) alloy. Nickel dendrites (light in b and c) are surrounded by a matrix of nickel-cerium eutectic. (a) 25⳯. (b) 75⳯. (c)
250⳯. Source: Ref 8

Downloaded from [Link]

by Ben Gurion University of the Negev user
 Metallography: An Introduction / 13

Fig. 25 Typical imperfections observable using optical microscopy. (a) Shrinkage porosity in an aluminum alloy 5052 ingot. Note angularity. 50⳯. (b) Coarse primary CrAl7crystal
in aluminum alloy 7075 ingot. 100⳯. (c) Oxide stringer inclusion in a rolled aluminum alloy 1100 sheet. 250⳯. All as-polished. Source: Ref 8

High-temperature phase transformations fre- crystallization occur on such a fine scale that electron microscope (SEM) and in some cases
quently nucleate at grain boundaries. The grain- TEM is necessary; however, light-microscope by examination of replicas with the transmission
boundary structures can be discrete or continu- metallography can be readily used to study most electron microscope (TEM). The SEM images in
ous. Continuous grain-boundary constituents of the recrystallization. The size of the recrys- Fig. 28 show the distinctive microscopic features
(Fig. 26) provide easy fracture paths when the tallized grains depends on the amount of cold of the three basics types of overload fracture:
grain-boundary phase is less ductile than the ma- working of the specimen before the recrystalli- transgranular brittle fracture (cleavage), trans-
trix phase. For the material shown in Fig. 26, the zation anneal. The greater the amount of cold granular ductile fracture (microvoid coales-
expected failure would be fracture along the work, the finer the recrystallized grain size (Fig. cence), and brittle fracture by intergranular sep-
grain-boundary carbides. Heterogeneous pre- 27). Because grain boundaries are a crystalline aration. The SEM provides good depth of focus
cipitation at grain-boundary regions is typically defect, continued annealing will cause this array to observe topological features of the fracture
based on the classic mechanism of precipitate of grains to be unstable, and grain growth will surface. Modern SEM instruments also typically
nucleation and growth, where the initial nucleus take place. Grain growth in a recrystallized spec- have x-ray spectroscopic attachments that allow
starts at critical size to allow reduction in the imen decreases the grain-boundary surface area elemental analysis of constituents on (or near)
interfacial surface energy between the precipitate to specimen volume ratio because the average the specimen surface. This can be very helpful
and parent phases (see the article “Structures by grain size increases as grain growth takes place. in failure analysis.
Precipitation from Solid Solution” in this Vol- The rate of grain growth depends on temperature However, lack of access to a SEM or TEM
ume). The transformation processes may also be and time. should not be viewed as a crippling obstacle to
continuous (e.g., see the article “Spinodal Trans- performing failure analysis, because such work
formation Structures” in this Volume). was done successfully prior to the development
Deformation Structures. The microscopic Microfractography
of these instruments. In many studies, such
details of deformation structures typically cannot (Adapted from Ref 13) equipment is not needed, while in other cases,
be fully established using light metallography.
Microscopic examination of the fracture sur- they are very important tools. In most cases,
Deformation changes the number and arrange-
face is best accomplished by use of the scanning electron microscopy and light microscopy
ment of dislocations (crystal defects) in the metal
should be considered complementary tools. Mi-
on an atomic scale. This dislocation substructure
crostructural examination can be performed with
is best characterized using TEM. Light-micro-
scope metallography can be used to supplement
TEM through characterization of the grain size
and anisotropy in grain shape and distribution.
Microstructural changes due to annealing can be
studied using TEM or light microscopy. The
most important structural changes that occur
during annealing are recovery, recrystallization,
and grain growth.
Recovery is the rearrangement and annihila-
tion of imperfections (primarily vacancies and
interstitials) within each grain of a cold-worked
polycrystalline component. Because recovery
deals mainly with point defects, any microstruc-
tural observations of it are difficult, and light mi-
croscopy cannot be used because of its limited
resolution.
Recrystallization is the formation of new
strain-free grains within the previously cold-
Fig. 26 Continuous grain-boundary precipitate in U-
700 nickel-base heat-resistant alloy. Etched us- Fig. 27 The effect of prior cold work on recrystallized
worked (strained) grains. The initial stages of re- ing HCl, ethanol, and H2O2. 500⳯. Source: Ref 8 grain size. Source: Ref 8

Downloaded from [Link]

by Ben Gurion University of the Negev user
 14 / Introduction

the SEM over the same magnification range as perpendicularly to the optical axis. It is best to it may be best to nickel plate the fracture to en-
the light microscope (LM), but examination with start with a low-power objective; long-working- hance edge retention. This is not required if the
the latter is more efficient. Contrast mechanisms distance types are preferred. Focusing reveals crack has not separated the component into two
for viewing microstructures are different for LM the limitations of the method, because only part pieces, or if a secondary crack is to be examined.
and SEM. Many microstructures, for example, of the fracture is in focus at any setting. Thus, Examination of the crack path using cross sec-
tempered martensite, exhibit poor contrast in the photographs* reveal only a portion of the frac- tions is also very useful for study of fractures
SEM and are best viewed by light microscopy. ture in focus, depending on the coarseness and due to environmental problems. Figure 31 shows
When atomic number contrast or topographic orientation of the fracture facets. Figure 29 a stress-corrosion crack in a partially broken
contrast is strong, the SEM provides good struc- shows an example of a brittle fracture in a low- sample of solution-annealed AISI 304 stainless
tural images, particularly above 500⳯ (Ref 14). carbon steel examined in this manner. Figure 29 steel tested in boiling (151 C, or 304 F) mag-
Again, because of the limitations and advantages also shows a LM image of the fracture profile, a nesium chloride. The crack path is predomi-
of each instrument, they are complementary LM image of a replica of the fracture, SEM im- nantly intergranular, but considerable transgran-
rather than competitive tools. All studies of mi- ages of the fracture and a replica of the fracture, ular fracture is also present. The SEM
crostructures and fractures should begin at the and a TEM replica of the fracture. Although fractograph of the specimen clearly reveals the
lowest magnification level, the unaided human Zapffe used bright-field illumination for this intergranular nature of the crack.
eye, and progress upward, first using the stere- work, dark-field illumination often produces su- Fracture profile examination is also very use-
omicroscope for fractures and the LM for frac- perior results. Figure 30 illustrates the use of ful in the study of failures due to liquid metal
ture path and microstructural studies, before us- bright-field and dark-field illumination for view- embrittlement (LME). Figure 32 shows the mi-
ing electron metallographic equipment. ing a brittle fracture in an Fe-Cr-Al alloy, plus a crostructure adjacent to a LME crack in a eutec-
Microfractography is a relatively new field; its SEM fractograph of the same area. Dark-field toid steel where liquid copper has penetrated the
roots can be traced to the light optical fracto- illumination is better at collecting the light scat- grain boundaries at 1100 C (2012 F) while the
graphs published by Zapffe and coworkers (Ref
tered from the fracture features; glare is reduced, sample was austenitic and under an applied ten-
15) beginning in the early 1940s, although a few
and image contrast is improved. Examination of sile load. Light microscope examination reveals
studies of historical value predated their efforts.
fracture replicas with the light microscope (Ref a discontinuous film of copper in the prior-aus-
Zapffe’s work, however, was almost exclusively
16–18) can extend the use of the method only to tenite grain boundaries and an intergranular frac-
confined to observation of cleavage facets on
rather brittle, coarse-grained specimens. The a limited extent, because the replica collapses ture path. Scanning electron microscope exami-
technique, basically an interesting academic ex- slightly, producing less depth of field. Also, with nation of the fracture also reveals the
ercise, did stimulate interest in fracture exami- a replica, the risk of damaging the objective is intergranular nature of the crack path.
nation as part of failure analysis. However, the eliminated. Surface detail can also be studied by LM using
depth-of-field limitation of the light microscope Considerable information concerning the frac- taper sections (Ref 27–29). This method has
has restricted its use for such work. Aside from ture mode and the relationship of the microstruc- been used to study wear phenomena, surface
the published light optical fractographs made by ture to the fracture path can be obtained by LM coatings, fatigue damage, and other fine surface
Zapffe (see Ref 16 for a review of many of examination of the profile of partially fractured detail. In this method, the surface is sectioned at
these), very few optical fractographs of metallic (Ref 19–21) or completely fractured (Ref 22–26) a slight angle to the surface. Polishing on this
materials have been published by others. Mi- polished metallographic specimens. Such ex- plane produces a magnified view of the structure
crofractography gained momentum with the de- aminations have been conducted for many years, in the vertical direction. The degree of magnifi-
velopment of TEM replication methods and be- long before the development of electron metal- cation is defined by the cosecant of the section-
came commonplace after the commercial lographic techniques, and continue to be used
introduction of the SEM in approximately 1965. because of the value of the method. If the frac- *With digital imaging, forging of the fracture surface can be
A flat, brittle fracture can be examined with ture has progressed to complete rupture, so that focused, and then all images combined to form one large image
the light microscope by orienting the fracture only one side of the fracture is to be examined, in focus.

Fig. 28 Scanning electron micrograph images of the basic types of overload fracture. (a) Intergranular fracture in ion-nitrided layer of ductile iron (ASTM 80-55-06). (b) Transgranular
fracture by cleavage in ductile iron (ASTM 80-55-06). (c) Ductile fracture with equiaxed dimples from microvoid coalescence around graphite nodules in a ductile iron
(ASTM 65-40-10). Picture widths are approximately 0.2 mm (0.008 in.) from original magnifications of 500⳯. Courtesy of Mohan Chaudhari, Columbus Metallurgical Services

Downloaded from [Link]

by Ben Gurion University of the Negev user
 Metallography: An Introduction / 15

ing angle; an angle of 5 43⬘ produces a tenfold spatial relationships, and shapes and to derive cent years, however, increased use has been
magnification. numerical data from micrographs. Some exam- made of manual and automated stereological re-
Finally, considerable progress has been made ples of image analysis include: lationships to describe microstructural character-
in applying the principles of quantitative metal- istics. Stereological methods are described in
● Quantitative determination of grain size,
lography to the study of fractures (Ref 25, 26, more detail in the article “Quantitative Charac-
grain shape, grain-boundary area per unit vol-
30–34). Much of this work has used measure- terization and Representation of Global Micro-
ume, and so on, in single-phase metals and
ments made on polished sections taken parallel structural Geometry” in this Volume.
ceramics
to the crack-growth direction (“vertical sec- The tedious nature of manual methods of such
● Quantitative determination of second-phase
tions”). This work provides new insight into measurements has spurred development of au-
volume fractions, sizes, interfacial areas per
fracture processes and should be useful in failure tomated procedures. Although microstructural
unit volume, spacings, and so on, in multi-
analysis, although its application to date has patterns, spatial relationships, and shapes are
phase metals and ceramics
been limited mainly to research studies. relatively easy to recognize in an image, reliable
● Quantitative determination of particle size
numerical data, such as counts, are far more dif-
distributions in powders
ficult to obtain manually. Therefore, automation
Historically, most microstructural ratings, par- addresses a long-recognized need in microstruc-
Image Analysis ticularly in quality-control studies and for spec- tural analysis for more precise data for quality-
(Adapted from Ref 37) ification compliance, have been performed using control and structure-property studies. The first
simple chart comparison ratings for features such approach was simply to develop devices to fa-
Metallography involves various methods to as grain-size measurements (e.g., ASTM E 112) cilitate manual data collection. These methods
compare and quantify microstructural patterns, and inclusion ratings (e.g., ASTM E 45). In re- reduce the influence of operator fatigue, which

Fig. 29 Cleavage fracture in a quenched-and-tempered low-carbon steel examined using three direct methods and three replication methods. (a) Light microscopy (LM) cross section
(nickel plated). Etched with Vilella’s reagent. (b) LM fractograph (direct). (c) Scanning electron microscopy (SEM) fractograph (direct). (d) LM replica. (e) SEM replica. (f)
Transmission electron microscopy replica. Source: Ref 13

Fig. 30 Light microscope fractographs taken with (a) bright-field and (b) dark-field illumination compared to (c) a scanning-electron secondary-electron image fractograph of the
same area. Sample is Fe-Al-Cr alloy. Source: Ref 13

Downloaded from [Link]

by Ben Gurion University of the Negev user
 16 / Introduction

affects the accuracy and reproducibility of man- NV, based on planar grain-size measurements Choice of magnification is basically a com-
ual measurements. and grain-shape assumptions. The further the promise. Small features require a relatively high
More recently, various semiautomatic and grains differ from the assumed shape, the greater magnification for accurate measurements, par-
fully automatic devices have been developed the error in estimating NV. ticularly for size, shape, or perimeter measure-
that permit more rapid data collection, analysis, When a relatively small polished area of one ments. However, the field area decreases with
and formatting. The continued improvements in or more samples is used to determine some quan- increasing magnification. Increasing the magni-
the speed and memory of affordable computers tity for a relatively large mass of material, errors fication improves resolution, but the area mea-
also have greatly aided the development of these also occur if sampling is inadequate or does not sured, if the number of fields is limited, may not
tools. The initial development of these devices represent the mass. This problem is common in be representative. For example, if there is a low
concentrated on hardware-centered systems, but analyzing inclusions in a heat of steel. Because volume fraction of inclusions that are not uni-
has since evolved to software-centered systems the quantity of inclusions is relatively or ex- formly distributed, a substantial area must be ex-
using faster, inexpensive computers with more tremely low, and the inclusion distribution is not amined to obtain a workable estimate of the vol-
memory. Thus, better-quality data can be ob- homogeneous, some degree of uncertainty is as- ume fraction. For such a sample, field-to-field
tained from analyzing a larger sample area and/ sociated with such measurements. This error can volume fraction variation increases as the mag-
or more samples. This allows better statistical be minimized by using a systematic sampling nification increases and thus results in a larger
accuracy and more meaningful results from im- plan and by increasing the number of samples standard deviation. One method of assessing this
age analysis. In addition, image analyzers can and area measured. In practice, compromise is is to examine the range of volume fraction mea-
perform several measurements on a field within necessary between the amount of time available surements as a function of the number of fields
milliseconds, providing a more complete de- for such measurements and the desired accuracy. for each magnification.
scription of the microstructure. Finally, auto- Effect of Magnification. The selected mag- The magnification effect has been observed in
mation can be introduced in stage movement, fo- nification influences the measurement value and many studies involving various materials, mea-
cusing, data analysis, and formatting. Image the statistics of the measurement. Numerous surements, and measuring devices. The problem
analyzers are available at various levels of so- cases have been documented in which the mea- is not simply one of poor technique. Two basic
phistication and cost, but all use electronic data sured values were altered significantly as the types of magnification-affected data can be
processing for image detection and measurement magnification was changed. Examples of this found in the literature. The first involves count-
of stereological and nonstereological parame- problem have been summarized (Ref 35). ing type data in which the count increases rap-
ters.
However, not all structures lend themselves to
accurate automatic detection; the structure must
exhibit adequate contrast to allow the analyzer
to distinguish its various components. For low-
contrast samples, semiautomatic analyzers may
provide more reliable feature discrimination at
the loss of some measurement speed. Sample
preparation procedures used for qualitative as-
sessment or manual measurements also may be
unsuitable for image analysis. Therefore, much
more care must be exercised in sample prepara-
tion. The ability to prepare the sample properly
is often the most critical and most difficult factor
in image analysis.
Obtaining maximum value from image anal-
ysis usually necessitates knowledge of sample
preparation, stereology, machine operation, sta-
tistics, and computer programming. Image anal-
ysis consists of sample selection and preparation,
image preprocessing, measurement, and data Fig. 31 Light micrograph of a cross section of (a) partially broken specimen and (b) a scanning electron fractograph
analysis and output. Each step must be con- of a completely broken specimen of solution-annealed AISI 304 stainless steel after stress-corrosion crack
trolled properly to obtain accurate, reproducible testing in boiling (151 C, or 304 F) magnesium chloride. Source: Ref 13
results. Sample selection must be systematic and
well planned to ensure that the samples analyzed
are representative. Image preprocessing refers to
the manipulation of the detected image to im-
prove the accuracy of measurements, for exam-
ple, separating adjoining particles before count-
ing, or to facilitate desired measurements, as in
the fusing of aligned inclusion stringers for a
length measurement.
Errors can arise in image analysis measure-
ments from such sources as the representative-
ness of the sample, quality of sample prepara-
tion, operator bias in setting controls, and
instrument errors. Errors can also result when
assumptions upon which stereological formulas
are based are invalid. A typical example of such
a problem is the mathematical procedure for de- Fig. 32 Light micrograph of (a) partially broken eutectoid carbon steel specimen embrittled by liquid copper at 1100
C (2012 F) (arrows point to grain-boundary copper penetration) and (b) scanning electron fractograph of the
termining the number of grains per unit volume, completely broken specimen. Source: Ref 13

Downloaded from [Link]

by Ben Gurion University of the Negev user
 Metallography: An Introduction / 17

idly with initial increases in magnification, then increases. For volume-fraction measurements, as crimination is a significant source of instrument
levels off at high magnifications. The second in- the volume fraction decreases, more fields must error (Ref 38). Machine problems may also in-
volves spacing measurements in which there is be measured to obtain the desired degree of ac- fluence measurement accuracy. Noise or random
an initial rapid drop in the spacing measurements curacy. This problem becomes particularly acute variations of current or voltage may also intro-
with an increase in magnification, followed by a when measuring volume fractions less than 0.01 duce errors. Noise problems become important
leveling off with further increases in magnifica- (1%). when measuring features with low contrast.
tion. Both situations arise from the observer’s A serious problem is encountered when the As previously noted, the quality and integrity
seeing more at the higher magnifications feature to be measured exhibits a rather wide of the specimens that are to be evaluated are
(Ref 35). range of sizes or a bimodal size distribution. The probably the most critical factors in image anal-
A somewhat different magnification problem larger particles are best measured at relatively ysis. No system can compensate for poorly pre-
may also be encountered in performing fully au- low magnifications, but the smaller particles pared specimens. While current systems can per-
tomated measurements. To illustrate this prob- must be measured at a higher magnification. At form gray image transformations at very rapid
lem, a medium-carbon hot-rolled steel contain- a high magnification, the guard-frame procedure rates relative to systems manufactured in the
ing ferrite and pearlite was analyzed for the must be used, or any large feature not within the 1970s and 1980s, gray image processing should
volume fraction of ferrite using 8, 16, 32, 50, 80, measurement area will be truncated and under- be used as a last resort for metallographic spec-
and 160⳯ objectives. The sample was etched sized. Two magnifications have been used for imens. Different etches and filters (other than the
using picral. At low magnifications, the volume such measurements, and the analyses have been standard green filter used in metallography)
fraction of proeutectoid ferrite could be mea- combined; however, this is a difficult procedure. should be evaluated prior to using gray image
sured with favorable accuracy. However, as the Sample preparation also influences mea- transformations. To obtain meaningful results,
magnification was increased, the structure of the surement accuracy significantly. Samples must the best possible procedures for polishing and
pearlite became resolvable, and ferrite within the be polished with minimum relief, the desired etching the specimens to be observed must be
pearlite constituent was also detected. Thus, with constituents must be retained, and polishing ar- used. Factors such as inclusion pullout, comet
increasing magnification and resolution, the vol- tifacts must be controlled. Because sample vol- tails, or poor or variable contrast caused by im-
ume fraction of ferrite increased. Figure 33 ume can be high, automatic polishing equipment proper etching cannot be eliminated by the im-
shows the measurement results as a function of is generally necessary to provide the required age analysis system. There is no substitute for
the magnification and number of fields mea- sample throughput and quality. Etching tech- properly prepared metallographic specimens.
sured. The best determination of the volume niques used for qualitative structure assessment One of the most common hardware variables
fraction of proeutectoid ferrite was obtained us- and manual measurements are frequently inade- that can affect system performance is the illu-
ing the 16⳯ objective. This is not obvious from quate for image analysis. Instead, selective etch- mination system of a light microscope. Proper
the data, but can be discerned while viewing the ing or staining techniques usually must be used. alignment of the microscope lamp and correct
detection. Color (tint) etching is extremely useful because adjustment are of paramount importance for op-
From a statistical viewpoint, lower magnifi- of its high selectivity and near absence of etch timal performance of the image analysis system.
cations enable measurement of large areas, less- relief. The alignment, resolution, and response of the
ening the influence of sample heterogeneity. As Electrolytic etching techniques are also valu- CCD camera used to convert the image into an
magnification increases, greater field-to-field able. The optimal procedure darkens either the electrical signal can have a large influence on
measurement variations are encountered. The phase of interest or all other phases, distinguish- performance. After properly aligning the micro-
choice of magnification influences the number of ing the constituents clearly by gray levels. In scope, some minor variations in the gray level of
fields to be measured. As magnification in- many studies, proper sample selection and care- a blank image might be observed. Because nei-
creases, more fields are required to obtain a cer- ful polishing and etching are the most critical ther the CCD camera nor the lenses in the mi-
tain degree of accuracy. Automatic stage move- factors in obtaining favorable results. This is es- croscope are perfect, some slight variations in
ment and automatic focusing facilitate pecially important in grain-size measurements, the gray image can occur. The shading corrector
increasing the number of fields. Automatic stage in which all the grain boundaries must be re- observes each pixel in a blank field of view and
movement ensures selection of fields without in- vealed clearly. changes its gray level to produce a uniform white
troducing operator bias. Automatic focusing Gray-Level Thresholding (Ref 36, 37). background. This correction factor is then ap-
should not be taken for granted. At higher mag- Many experiments have demonstrated that the plied to every image to correct for minor prob-
nifications, the potential for improper focusing setting of the detector threshold for feature dis- lems in the system. If a large degree of correction

Fig. 33 Volume fraction of ferrite as a function of number of fields measured and magnification. At high magnifications, equiaxed ferrite within the coarse pearlite were detected.
Source: Ref 37

Downloaded from [Link]

by Ben Gurion University of the Negev user
 18 / Introduction

is required, this may indicate that the microscope ture detection, five iron-carbon alloys with car- 36(b), the threshold has been set to detect only
is seriously misaligned, or the CCD camera has bon contents of 0.003 and approximately 0.2, the pale gray sulfides that appear white on the
deteriorated from use. 0.4, 0.6, and 0.8% were prepared metallograph- screen. Detection setting can be aided by alter-
Once these concerns have been properly ra- ically and etched using picral. The microstruc- nating between the live image and the detected
tionalized, the next major variable to consider is tures consisted of varying amounts of ferrite and image (flicker mode) while observing the size
how to select the gray level of the objects to be pearlite. The samples were scanned in 1% incre- correlation between these two images. If the
characterized as described further in Ref 36. For ments from black to whiter. gray-level ranges of two constituents overlap,
optimal detection accuracy, the sample features Figure 34 (Ref 37) shows the area percent in detected (white) points will be visible in the un-
of interest must be treated to have as narrow a the detected field area at each 1% portion in the desired phase as the threshold setting is changed.
contrast (gray-level) range as possible (Ref 37). gray-level scan for these alloys. The 0.003% C If the degree of overlap is excessive, an alternate
The light source in the microscope must then be alloy, consisting almost entirely of ferrite, exhib- preparation procedure is required. In nearly all
aligned for even illumination. Most image ana- its the highest ferrite peak, with maximum de- inclusion studies, the sample is examined un-
lyzers provide shading correction to even out tection at approximately 81% on the threshold etched because detection of the inclusions is
variations across the screen. The gray-level device. As the carbon content increased to 0.6%, more reliable. Figure 36(c) shows detection of
range of the image from the darkest black to the the size of the ferrite peak decreased, and the the martensite; Fig. 36(d) detection of the d-
lightest white feature is usually segmented into peak position increased slightly. For the 0.8% C ferrite. The optimal procedure for d-ferrite de-
256 increments. As an example of gray-level fea- alloy (all pearlite), no image was detectable tection is to etch the sample electrolytically us-
above 76%. This suggests that the pearlite con- ing 20% aqueous NaOH, which colors only the
stituent is generally detected between approxi- d-ferrite.
mately 16 and 76%, and the ferrite constituent is In attempting to detect all three constituents
detected between approximately 76 and 91 to in this manner, it may be difficult, if not impos-
99%, depending on the alloy. Figure 35 shows sible, to have the three volume-fraction mea-
the same data plotted as a cumulative percentage surements add up to 100% on every measure-
of area fraction. ment field. This example illustrates the ability to
Setting the threshold to detect only gray levels separate a complex image by thresholding. The
within specific ranges enables selective detection preferred procedure would be to measure the sul-
of constituents. For example, Fig. 36 shows the fides in the as-polished condition, etch electro-
microstructure of AISI type 416 stainless steel lytically and measure the amount of d-ferrite,
etched using Vilella’s reagent. This is a resulfur- then determine the amount of martensite (major
ized grade that has been heat treated to form tem- constituent) by difference.
pered martensite but also contains d-ferrite Figure 37 illustrates the degree of error that
stringers. Figure 36(a) shows the live micro- can be encountered when thresholding is incor-
scope image before phase detection. In Fig. rect. Figure 37(a) shows the live image of a

Fig. 34 Plot of the detected area fraction in 1% incre-

ments from black to white for five iron-carbon
alloys. Microstructures consisted of varying amounts of fer-
rite and pearlite, ranging from the 0.003% C alloy (almost
all ferrite) to the 0.8% C alloy (all pearlite). Source: Ref 37

Fig. 35 Plot of cumulative detected area fraction in 1% Fig. 36 Examples of preferential detection in an AISI 416 stainless steel sample. (a) Live image. (b) Preferential detec-
increments from black to white for the five tion of manganese sulfides (white). (c) Preferential detection of tempered martensite (white). (d) Preferential
iron-carbon alloys in Fig. 34. Source: Ref 37 detection of d-ferrite (white). Sample etched using Vilella’s reagent. 175⳯. Source: Ref 37

Downloaded from [Link]

by Ben Gurion University of the Negev user
 Metallography: An Introduction / 19

ferrite-pearlite sample etched using picral. Picral gion to prevent particle measurement errors. The per unit area, the image analyzer must first de-
is far more accurate for such measurements be- region between the edge of the screen and the termine what portions of the detected image are
cause it does not etch the ferrite grain bound- live frame is the guard region. discrete particles. A particle is any detected fea-
aries, which would be detected with the pearlite, Image analyzers can store several images in ture completely surrounded by undetected pic-
and it produces more uniform darkening of the memory and use these images variously to en- ture points. After this discrimination, the parti-
pearlite than nital. Figure 37(b) shows an under- hance feature selection. The ability to erode or cles are counted, and the number is divided by
detected image containing 28.1% pearlite. Some dilate features and compare these images to the the measurement area.
of the pearlite patches contain undetected re- original image can be useful in separating con- Sizing can be performed in several ways. The
gions. Figure 37(c) illustrates a correctly de- tiguous particles or fusing stringered features. simplest procedure for determining the average
tected field with 34.05% pearlite. Figure 37(d) Image analysis is much more precise than man- particle area, Aavg, is to divide the area fraction
shows excessive detection, in which the features ual techniques for the counting of particles, but by the number of particles per unit area, that is,
are enlarged and the volume fraction of pearlite adjoining particles are difficult to handle without AA/NA. This is the most direct procedure using
is 42.3%. these special image-editing procedures. Accurate field measurements. Feature-specific measure-
The above examples illustrate field measure- counting is also influenced by the shape of a par- ments provide several possible procedures. One
ments; that is, all the features are measured si- ticle and the counting procedure used. Spherical is to measure the area of each feature and deter-
multaneously. Image analyzers also can perform particles are easiest to count. More complex mine the average area of all the measured par-
feature-specific measurements; that is, each dis- shapes require selection of the proper counting ticles. Another involves measuring the area of
tinct feature in the field is measured individually. technique to obtain accurate counts. each particle and computing the equivalent di-
This technique is highly useful for particle sizing Measurements are performed based on the ameter based on some assumption about their
or shape measurements. When such capability is number and distribution of detected picture shape. The projected height of the particles, the
available, features with the same gray level can points relative to the scan line and the total num- maximum and minimum length and thickness,
be further discriminated by differences in size or ber of picture points. For example, the area frac- and so on, could also be measured.
shape. For example, the various graphite shapes tion is simply the ratio of the number of detected
of a cast iron sample can be sorted by size or picture points to the total number of picture ACKNOWLEDGMENT
shape, then measured. points in the measurement field; that is, the de-
Most image analyzers can perform measure- tected area divided by the measurement area. This article contains excepts and adapted text
ments over the full screen (blank frame) or Therefore, the true size of the measurement field from the following sources:
within a smaller frame. In Fig. 36 and 37, the must be determined accurately using a stage mi-
● W.T. Becker, Fracture Appearance and Mech-
central screen area within the vertical and hori- crometer for each magnification to operate in the
anisms of Deformation and Fracture, Failure
zontal lines is the live-frame region, which is calibrated mode. Other measurements are more
Analysis and Prevention, Volume 11, ASM
generally used when feature-specific measure- complex. For example, to count the number of
Handbook, 2002, p 559–586
ments are conducted. Particles that intersect the detected particles in an image field of known
● B.L Bramfitt and A.O. Benscoter, Metallog-
live frame can be deleted from the detected re- size, that is, to obtain NA, the number of particles
rapher’s Guide: Practices and Procedures for
Irons and Steels, ASM International, 2002, p
88
N
● D.W. Hetzner, Applications, Chapter 8 in
Practical Guide to Image Analysis, ASM In-
ternational, 2000, p 203–256
● M.R. Louthan, Jr., Optical Metallography,
Materials Characterization, Volume 10, ASM
Handbook, American Society for Metals,
1986, p 299–308
● R.A. Lund and S. Sheybany, Fatigue Fracture
Appearances, Failure Analysis and Preven-
tion, Volume 11, ASM Handbook, 2002, p
627–640
● L.E. Samuels, Metallographic Polishing by
Mechanical Methods, 4th ed., ASM Interna-
tional, 2003, p 6
● G.F. Vander Voort, Image Analysis, Materials
Characterization, Volume 10, ASM Hand-
book, American Society for Metals, 1986, p
309–322
● G.F. Vander Voort, Metallographic Tech-
niques in Failure Analysis, Failure Analysis
and Prevention, Volume 11, ASM Handbook,
ASM International, 2002, p 498–510

REFERENCES
1. H.C. Sorby, in How to Work with the Mi-
croscope, London, 1867
2. C.S. Smith, A History of Metallography,
University of Chicago Press, 1960
Fig. 37 Example of the influence of detection setting on the area fraction (AA) of pearlite detected in a low-carbon
3. H.C. Sorby, J. Iron Steel Inst., Vol 28, 1886,
steel. (a) Live image showing ferrite and pearlite. (b) Pearlite underdetected (AA ⳱ 28.1%). (c) Pearlite detected
correctly (AA ⳱ 34.05%). (d) Pearlite overdetected (AA ⳱ 42.3%). Sample etched using 4% picral. 180⳯. Source: Ref 37 p 140

Downloaded from [Link]

by Ben Gurion University of the Negev user
 20 / Introduction

4. B.L. Bramfitt and A.O. Benscoter, Metal- 15. C.A. Zapffe and M. Clogg, Trans. ASM, Vol 29. M.H. Hurdus, “Taper Sectioning of Tubular
lographer’s Guide: Practices and Proce- 34, 1945, p 71–107 Specimens and Its Application to Corrosion
dures for Irons and Steels, ASM Interna- 16. K. Kornfeld, Met. Prog., Vol 77, Jan 1960, Oxide Film Examination,” Report AERE-
tional, 2002, p 88 p 131–132 R9704, U.K. Atomic Energy Authority,
5. H. Wedding, J. Iron Steel Inst., Vol 27, 1885, 17. P.J.E. Forsyth and D.A. Ryder, Metallurgia, Harwell, Oct 1980
p 187 March 1961, p 117–124 30. S.M. El-Soudani, Metallography, Vol 11,
6. L.E. Samuels, Metallographic Polishing by 18. K.R.L. Thompson and A.J. Sedriks, J. Aust. July 1978, p 247–336
Mechanical Means, 4th ed., ASM Interna- Inst. Met., Vol 9, Nov 1964, p 269–271 31. E.E. Underwood and E.A. Starke, Jr.,
tional, 2003, p 6 19. H.C. Rogers, Trans. AIME, Vol 218, June “Quantitative Stereological Methods for
7. J.R. Vilella, Metallographic Techniques for 1960, p 498–506 Analyzing Important Features in Fatigue of
Steels, American Society for Metals, 1938 20. C. Laird and G.C. Smith, Philos. Mag., Vol Metals and Alloys,” STP 675, Annual Book
8. M.R. Louthan, Jr., Optical Metallography, 7, 1962, p 847–857 of ASTM Standards, ASTM, 1979, p 633–
Materials Characterization, Vol 10, ASM 21. R.H. Van Stone and T.B. Box, “Use of Frac- 682
Handbook, American Society for Metals, tography and Sectioning Techniques to 32. E.E. Underwood and S.B. Chakrabortty,
1986, p 299–308 Study Fracture Mechanisms,” STP 600, An- “Quantitative Fractography of a Fatigued
9. W.T. Becker, Fracture Appearance and nual Book of ASTM Standards, ASTM, Ti-28V Alloy,” STP 733, Annual Book of
Mechanisms of Deformation and Fracture, 1976, p 5–29 ASTM Standards, ASTM, 1981, p 337–354
Failure Analysis and Prevention, Vol 11, 22. W. Staehle et al., Corrosion, Vol 15, July 33. M. Coster and J.L. Chermant, Int. Met. Rev.,
ASM Handbook, ASM International, 2002, 1959, p 51–59 (373t–381t) Vol 28, 1983, p 228–250
p 559–586 23. D. Eylon and W.R. Kerr, “Fractographic and 34. E.E. Underwood, Quantitative Fractogra-
10. R.W. Hertzberg, Deformation and Fracture Metallographic Morphology of Fatigue Ini- phy, Applied Metallography, Van Nostrand
Mechanics of Engineering Materials, 4th tiation Sites,” STP 645, Annual Book of Reinhold, 1986, p 101–122
ed., John Wiley, 1996 ASTM Standards, ASTM, 1978, p 235– 35. E.E. Underwood, “Practical Solutions to
11. R.A. Lund and S. Sheybany, Fatigue Frac- 248 Stereological Problems,” STP 839, Annual
ture Appearances, Failure Analysis and Pre- 24. W.R. Kerr et al., Metall. Trans., Vol 7A, Book of ASTM Standards, ASTM, 1984, p
vention, Vol 11, ASM Handbook, ASM In- Sept 1976, p 1477–1480 160–179
ternational, 2002, p 627–640 25. W.T. Shieh, Metall. Trans., Vol 5, May 36. D.W. Hetzner, Applications, Chapter 8 in
12. Fatigue Failures, Failure Analysis and Pre- 1974, p 1069–1085 Practical Guide to Image Analysis, ASM In-
vention, Vol 11, ASM Handbook, ASM In- 26. J.R. Pickens and J. Gurland, Metallographic ternational, p 203–256
ternational, 2002, p 708 Characterization of Fracture Surface Pro- 37. G.F. Vander Voort, Image Analysis, Mate-
13. G.F. Vander Voort, Metallographic Tech- files on Sectioning Planes, Proc. Fourth In- rials Characterization, Volume 10, ASM
niques in Failure Analysis, Failure Analysis ternational Congress for Stereology, NBS Handbook, American Society for Metals,
and Prevention, Vol 11, ASM Handbook, Spec. Publ. 431, 1976, p 269–272 1986, p 309–322
ASM International, 2002, p 498–510 27. E. Rabinowicz, Met. Ind., Vol 76, 3 Feb 38. C. Fisher, The Quantimet: Setting the
14. G.F. Vander Voort, The SEM as a Metallo- 1950, p 83–86 Threshold and the Correction of Off-Set
graphic Tool, Applied Metallography, Van 28. L.E. Samuels, Metallurgia, Vol 51, March Threshold Error, Pract. Metallogr., Vol 6,
Nostrand Reinhold Co., 1986, p 139–170 1955, p 161–162 Nov 1969, p 659–672

Downloaded from [Link]

by Ben Gurion University of the Negev user