Published by Buehler, a division of Illinois Tool Works

Volume 1, Issue 5

Introduction to Quantitative Metallography

By: George Vander Voort

Quantitative Metallography- 30 Years Ago I Discover Stereology

For most of its history, metallographic observations have been In the early 1970’s, I came across some articles and a book,
largely qualitative in nature. The structure might be described as Quantitative Stereology, written by the late Ervin E. Underwood
being relatively coarse or fine, or layered, or uniform. Particles might (Addison-Wesley Publishing Co., 1970). This was the beginning of
be referred to as being globular or spheroidal, lamellar, acicular, or a lifelong interest in quantifying microstructures and a long time
blocky. Microstructures were single-phase or duplex, and so forth. friendship with a fascinating gentleman, Erv Underwood.

Thirty years ago when I entered industry, chart ratings and visual
examinations were the main approach toward quantification. I can
well remember the mill metallographers looking at spheroidized
carbide tool steel structures and stating that it was, for example,
95%
spheroidized (many raters would never say 100%, just as some
teachers would never grade an essay at 100%!) or that it was 60%
spheroidized and 40% lamellar tending to spheroidize. Or, without
looking at the chart (a seasoned rater never did), they would
pronounce that the grain size was, for example, 100% 6 to 8 or
perhaps 70% 8 and 30% 3 to 5 if it was duplex in appearance.

As a novice metallurgist, I was quite impressed by these

pronouncements and tried to repeat the practice on my own. But, The late Erv Underwood at his home in Big Canoe, Georgia
I found myself unable to repeat such ratings, if done several weeks
apart, unless I made my estimates very broad. Later, I tried submitting
the same specimens to different metallographers or to the same Stereology is used to quantify matrix microstructures, as opposed to
person at different times. I quickly learned that their repeatability/ standard metrology techniques for measuring case depth, plating
reproducibility wasn’t that much better than mine. (Don’t ever let a thickness, or particle size. Microstructural measurements are made
mill metallographer know that you are checking him/her!) on a two-dimensional plane-of-polish through a three-dimensional
opaque metal. Stereology converts these 2-D measurements into 3-D
The greatest mystery to me, however, was inclusion chart ratings. I estimates of microstructural parameters. Most procedures are very
was very impressed that they could scan the 160 square mm area in a simple to use, but there are special considerations for their validity.
few minutes and, without taking any notes, jot down worst-field
ratings. Absolute black magic! I never could do this. I had to use a Metallographers commonly perform metrology type measurements;
scale and measure stringer lengths or count inclusions when I saw e.g., when measuring case depths, decarburization or plating/
a field that looked like it had a high inclusion content. And I had to coating thickness. A scale is placed over the structure, and the depth
keep notes. But, I was a young, “hot-shot” metallurgical engineer or thickness perpendicular to the surface is measured. For example,
– why couldn’t I do as well? Was I overly dense? Did I go to too ASTM E 1077 describes the measurement of decarburization of steel
many parties and too few classes? specimens using such measurements.

Naturally, I tried testing their ability to reproduce test results, and I Nomenclature
found out that maybe I wasn’t as slow as I thought. They didn’t do Application of stereology has been hampered by confusion due to
that well reproducing their own results. Later, when I became active the use of different mathematical symbols for the same parameters.
with ASTM Committee E-4, I found that inclusion ratings on round- To minimize this problem, the International Society for Stereology
robins, which had been tried numerous times, were notoriously has promoted a standard nomenclature which is constantly evolving
non-reproducible. Well, that was good for soothing my shattered as new approaches are developed. The most basic symbols are:
ego, but it did not solve my problems of describing structures.

Solutions for Materials Preparation, Testing and Analysis

 P = Point
L = Line
A = Area + + + + + + + +
S = Surface
V = Volume + + + + + + + +
N = Number + + + + + + + +
+ + + + + + + +
These symbols can be combined in a number of ways to generate
different symbols. For example, PP represents the point fraction; + + + + + + + +
that is, the fraction of grid points lying in a phase of interest. While
+ + + + + + + +
A and S seem to be the same, A is for a flat surface while S is for
a curved surface. Thus, SV represents the grain boundary surface + + + + + + + +
area per unit volume. NA is the number of particles per unit area + + + + + + + +
while NV is the number per unit volume.

Phase Proportions
One of the most common measurements, determination of the Figure 2. “Super-pure” aluminum electrolytically etched with Barker’s
reagent and viewed with polarized light plus a sensitive tint filter
amount of phases present, can be done using three different
revealing an equiaxed, recrystallized grain structure. There are some
methods. Areal analysis, developed by Delesse in 1848, says that fine impurity precipitates in this specimen, however, that cannot be
the area percent of a phase on a 2-D plane is equal to its volumetric seen with this etch and magnification (50x).
percent, that is, AA = VV. However, measuring the area of second
phases is very tedious unless they are quite coarse. Lineal analysis, The microstructure above shows the beta phase in Muntz metal
developed by Rosiwal in 1898, says that the lineal fraction of (Cu-40% Zn) preferentially colored by Klemm’s I reagent while the
test lines in a phase on the 2-D plane is equal to its volumetric alpha matrix is unaffected - ideal conditions for point counting. Since
percentage, that is, LL = VV. This is easier to determine but still there is less β than α, we will count the number of times the points
rather tedious. fall in the colored β grains. The amount of α is simply 100 - %β. As
you can see, we have superimposed a 64-point test grid (8 rows of
Starting around 1930, several workers in different fields and 8 points) over the structure and we have 15 hits and 4 tangent hits.
countries showed that the percentage of points on a test grid lying The point fraction (volume fraction) is 17/64 = 0.266 or 26.6%.
in the phase of interest was equal to the volumetric percentage,
that is, PP = VV. Of the three methods, this is the most efficient The point counting grid would be placed randomly over the structure
technique; that is, it produces the best precision for the least effort a number of times so that the point fraction is determined for a
when done manually. The point counting technique is described fully number of fields. The necessary number of fields to yield a 10%
in ASTM E 562 (also ISO 9042). Image analyzers use essentially the relative accuracy varies inversely with the volume fraction (the lower
same procedure; that is, the amount of a phase (usually called the the volume fraction, the greater the number of fields, i.e., the greater
area fraction or volume fraction even if it actually is a point fraction) the total number of applied grid points).
is determined by the number of picture elements or “pixels” in
the phase of interest divided by the total number of pixels; i.e., PP, Grain Size
expressed usually as a percentage. Grain size is perhaps the most commonly performed microstructural
measurement, although chart ratings are more commonly done
Point Counting Example than actual measurements (this is changing). A recent ASTM inter-
ASTM E 562 describes the point counting procedure for determining laboratory “round-robin” showed that chart ratings of grain size are
the amount of second-phase constituents. A grid with systematically biased; that is, the ASTM grain size number is 0.5 to 1 unit too low
spaced points (e.g., 10 rows of 10 equally spaced points) is (See Appendix X1 of E 112 -96). No bias existed when planimetric
superimposed over the structure, either on an eyepiece reticle or measurements were compared to intercept measurements by the
a plastic sheet placed over or behind a ground glass projection same raters.
screen or on a TV monitor screen. The points are usually drawn as
fine perpendicular crossing lines and the “point” is the intersection The ASTM grain size number, G , is defined as:
of the two lines. This is done because actual points would be very n = 2 G-1
difficult to see. The optimum point density for manual point counting
is usually determined from 3/VV where the volume fraction is a where n is the number of grains per square inch at 100X. To convert
fraction (not a percent). If the volume fraction is 0.5 (50%), then the n to NA (the number of grains per square mm at IX), multiply n by
optimum grid point density is 6. On the other hand, if the volume 15.5. The four ASTM grain size charts show graded series of grain
fraction is 0.01 (1%), the optimum point density is 300. The point structures of different types.
fraction is the ratio of the points in the phase of interest to the
number of grid points. Some people like to use a 100 point grid Grain size can be measured by the planimetric method (developed
for all work since the division is unnecessary. Points falling on the by Zay Jeffries in 1916) or by the intercept method (developed by
interface are counted as ½ a hit. For best manual results we need Emil Heyn in 1904). In the planimetric method, a count is made of
to sample more fields and do as little work as possible on each field the number of grains completely within a circle of known area and
measurement (the adage, “do more, less well”). The field-to-field half of the number of grains intersected by the circle to obtain NA.
variability has a greater influence on measurement precision than Then, NA is related to G. This method is slow when done manually
the counting precision on a given field. because the grains must be marked when counted to obtain an
accurate count.

Visit our website at www.buehler.com for more information. 2

In the intercept method, either straight lines, curved lines, or circles The ASTM grain size can also be determined using the intercept
are placed over the structure and a count is made of either the method counting either the number of grains intercepted, N, or the
number of grain boundary intersections, P, or the number of grains number of grain boundaries intersected, P, with a test line. ASTM
intercepted, N, by the line. P or N is divided by the true line length, LT, recommends using a grid with three concentric circles with a 500mm
to determine PL or NL, the number of intersections or interceptions total line length. To illustrate the principle of the method, we will
per unit length (for a single phased structure). The reciprocal of PL use the same image with a single circle (shown on previous page).
or NL gives the mean lineal intercept length, l, The count of the number of grains intercepted by the circle is N. To
l = 1/NL = 1/PL calculate the number of interceptions per mm, NL, we divide N by
a measure of grain size that can be converted to a G value. The the true length (circumference) of the circle. Since the diameter of
intercept method is more efficient than the planimetric method the circle is 64.4mm, its circumference is πD, that is, 202.3mm. The
yielding acceptable measurement precision (<10% relative accuracy) true length is 202.3mm divided by the magnification, M, that is,
in much less time. ASTM E 112 contains a complete description of 1.01mm. Hence, NL = N/LT = 34/ 1.01 = 33.6 interceptions per mm.
these methods. A major revision of E 112 was approved in 1995.
To calculate the grain size, we first determine the mean lineal
intercept length, l, which is the reciprocal of NL (or of PL, the number
Grain Size Example of grain boundary intersections per unit length). Thus, l = 1/ 33.6
ASTM E 112 describes the manual measurement of grain size for = 0.0298mm. G is calculated from an equation from E 112-96:
structures with a single grain size distribution while ASTM E 1382 G = {-6.644 (log10 l) – 3.288}
covers image analysis measurements. Note that E 112 was heavily where l is in mm. In this example, G = 6.85. Since the two methods
revised in 1995 (additional minor changes in 1996), so it is best to are sensing different geometric aspects of the three-dimensional
read the latest version. Grain size can be measured using either grain structure, they will not give exactly the same value, but they
will be close, generally within the experimental limitations of the
the planimetric or the intercept methods. In the examples, we will
measurements. In practice, we would repeat these measurements
simplify the approach slightly to illustrate the methods. Additional on a number of fields in order to obtain a good estimate of the
field sampling should be done to obtain good statistical data. grain size.

In the planimetric method, ASTM recommends using a test circle

with a diameter of 79.8 mm (5000 sq. mm area) placed randomly Spacings
over the grain structure. To obtain an accurate count of the number The spacing between second-phase particles, such as carbides or
of grains inside the circle and the number intercepted by the circle, inclusions in steels or between intermetallic particles in aluminum
we must mark the grains on the template as we count which makes alloys, can affect mechanical properties and formability. A special
this method slow (although this is not a problem by image analysis). case is the interlamellar spacing of pearlite in high carbon steels (like
rail steel) where refinement of the spacing improves both strength
We must know the magnification of the image as well.
and toughness.
The image below shows the grain structure at 200X of a lowcarbon Spacings are easily assessed using a simple NL (number of particles
sheet steel after color etching. A circle of known size (64.4mm intercepted per unit length of test line) measurement. The mean
diameter) has been placed over the image to illustrate the method. center-to-center spacing, sometimes called σ, is simply:
There are 44 grains within the circle (ninside) and 34 grains σ = 1/NL
intercepted by the circle (nintercepted). The number of grains per This is not a nearest-neighbor spacing, but the mean spacing
sq. mm, NA, is calculated from: between particles in the test line direction (either placed randomly or
NA = f {ninside + ½(nintercepted)} in some preferred direction, such as the through-thickness direction).
The multiplier f is calculated from (M2/circle area), where M is the
If the amount of the second phase is determined, for example, by
linear magnification of the image. For this example,
point counting, the mean edge-to-edge spacing, called λ (or the
NA = 12.28 {44 + ½(34)} = 749.1 grains /sq. mm mean free path, MFP), can be calculated by:
From NA, we can calculate the ASTM grain size number, G, using λ = (1-PP)/NL
the following formula from E 112-96: where PP is a fraction rather than a percentage. This is a very good
G = {3.322 (log10 NA) – 2.954} = 6.6 structure-sensitive parameter.

By a simple subtraction of (σ - λ), we can obtain the mean intercept

length of the second phase particles – without measuring any
particles! Furthermore, if we count the number of particles within
a known area to obtain NA (including only half of the particles
intersected by the field edges), we can determine the average cross
sectional area of the particles, A, by:
A = PP /NA
where PP is the point fraction (as a fraction, not a %) of the second
phase. Thus, the average size of particles can be determined
manually without actually measuring the particles. With modern
image analyzers, individual measurements of particles are fast
and simple Besides generating average particle dimensions, the
distribution of particle sizes can be obtained by feature-specific
image analysis.

To determine the interlamellar spacing of pearlite (or of any eutectic

or eutectoid), it is common practice to count the number of carbide
interceptions with a straight test line perpendicular to the lamellae.

Visit our website at www.buehler.com for more information. 3

However, because the lamellae intersect the surface at different A good measure of statistical precision is the 95% confidence
angles, it is better to determine a mean random spacing, σr, than interval (or confidence limit). This defines a range around the mean
a mean directed spacing, σd. A mean random spacing is obtained value where, 95 times out of 100, a subsequently determined mean
by determining NL using randomly oriented test lines (or curved or will fall. For example, a mean volume fraction of 10% ± 2% implies
circular lines). The mean random spacing is easily used to calculate that for 95 of 100 measurements, the mean value will be between
the mean true spacing, σt, by: 8 and 12%. The 95% confidence interval is determined by:
σt = σr /2 95% CI = ts/n½
In the past, the mean directed spacing, σd, was determined for where t is the Student’s t factor (t is a function of the confidence
the pearlite colony with the finest observed spacing, and this level desired and the number of measurements, n, and can be found
was assumed to be the true spacing. This is a better technique in standard textbooks and in some ASTM standards, e.g., E 562 and
for isothermally-formed pearlite than for pearlite formed during E 1382) and s is the standard deviation.
continuous cooling. However, the longer you search for the finest
colony, the finer the measured colony size! That is, the σd value The relative accuracy, RA, of a measurement is determined by:
obtained depends upon the amount of time spent looking for %RA = 100 · (95% CI)/X
the finest colony, even in isothermally-formed pearlite. Any effort where X is the mean value. In general, a relative accuracy of 10%
spent looking for a “best” or “worst” field condition, of any type, or less is considered to be satisfactory for most work.
is strongly influenced by the amount of search time, and the results
are neither reproducible nor precise. Sampling
So far, we have discussed measurements on a single plane-of-
Interlamellar Spacing Example polish on one specimen. Because we are usually dealing with large
Traditionally, the metallographer has searched for the finest quantities of material (such as an entire “heat” of metal or alloy, a
appearing interlamellar colony and made a measurement of its large heat treatment lot of bars, forgings, etc., or a large forging or
spacing using a test line perpendicular to the lamellae. This spacing casting), a single specimen may not be representative of the whole
is claimed to be the true interlamellar spacing. However, this method quantity. Ideally, random sampling of a large batch would be best,
is not reproducible as the longer you search, the finer the measured but practical considerations usually rule this out.

In most cases, sampling is done at predetermined convenient

locations, such as the extreme ends of a coil, bar, plate, etc., or at
locations which will be subjected to extensive machining. In some
cases, excess metal is added to a forging or casting to provide
test material as similar as possible to that of the component.
Sampling is often a compromise and is rarely excessive due to cost
considerations. However, inadequate sampling or sampling of
nonrepresentative material or locations may degrade the value of
the measurements.

Stereological measurements are best employed on sectioning

planes that sample the microstructure randomly. This means
that any oriented plane will produce the same data within the
limits of statistical precision. However, for certain materials, the
microstructure varies with the test plane. A classic example is that of
inclusions in wrought steels which are preferentially elongated in the
deformation direction. Sampled perpendicularly to the deformation
axis (transverse plane), the inclusions look like spherical particles
spacing. A better approach is to measure a mean random spacing while, when sampled on a plane parallel to the deformation axis
and divide that by two to get the mean true spacing. This method (longitudinal plane), they appear as long, thin rods or as broken
was verified and proven to be correct (see Metallography, Vol. 17, “stringers”. If measurements are made on these planes, we obtain
No. 1, February 1984, pp. 1-17). The micrograph shown above is different values for NA, their length (or diameter), their spacing, and
that of an as-rolled carbon steel of about 0.45% C, etched with 4% even the volume fraction. Thus, if we want to characterize their 3-D
picral and photographed with the SEM (specimen perpendicular to characteristics, measurements must be made on the three principal
the beam) using secondary electrons at 17,800x magnification. A planes and averaged (or an alternate technique such as the trisector
circle with a diameter of 49.7mm was placed over the lamellae and with vertical sectioning must be employed). In practice, the true 3-D
the number of carbide lamellae intercepted by the test line, N, was characteristics may not be needed, and measurements are made
counted. NL was again determined as N (23) divided by the true using one standard test plane orientation which yields data suitable
line length, πD/M, where the circle diameter, D, is 49.7mm and M is for quality control and material comparisons. This is the procedure
17,800x. Thus NL is 2622 interceptions per mm. The mean random employed in ASTM E 1245.
spacing, σr, is given by:
σr = 1/NL = 0.381μm = 381nm Specimen Preparation
The mean true spacing, σt, is σr /2 = 190.7nm. If we are to obtain precise, unbiased measurements, we must
prepare the specimens to reveal the true structure. In general, it is
Statistics best to prepare specimens so that only the desired aspect of the
Other measurements are possible, but the ones described above structure is visible. In this way, the eye sees only what is needed,
represent some of the simplest and most useful. Each can be and other information is not present to obscure what is of interest.
repeated on a number of fields on the plane-of-polish so that a To study inclusions, intermetallics, or other constituents with a
mean and standard deviation can be obtained. The number of reflectivity or color different than the matrix, etching is not required.
fields measured influences the precision of the measurement. To measure matrix phases, use a selective etch that reveals only the
Manual measurements are tedious and time-consuming so sampling constituent of interest. For example, for a carbon steel, etch with
statistics may be less than desired. Image analysis removes most of picral and point count the amount of ferrite or pearlite (always
the barriers to inadequate sampling. point count the constituent with the lower concentration). Etching

Visit our website at www.buehler.com for more information. 4

with nital will bring up the ferrite grain boundaries and produce grid density is a function of the volume fraction and is found using
less uniform etching of the pearlite making point counting more the equation 3/VV. Thus, for 5, 3 or 1% volume fractions (0.05,
difficult and less precise. 0.03 and 0.01, respectively), the optimum point densities are 60,
100 and 300 points, respectively. A grid with 500 points takes more
The laws of stereology for measuring microstructure are based on time to count than a lower point grid. For example, a 500 point grid
the assumption that the etched surface is planar. There are rules for takes five times as long to count as a 100 point grid. To optimize
non-planar (curved) surfaces and for measurements with transmitted the method, that is, to make it most efficient, we want to measure
light or electrons, but these are much more difficult to employ. as many fields as possible in the least amount of time (in agreement
For microstructural measurements, the surface must be flat and with the adage, “do more, less well”). To obtain a 10% relative
etch depth should be minimal (do not over-etch!). Deep etching accuracy with a 5% volume fraction and a 100 point grid, we do
can dramatically change the apparent amount of a second phase need to analyze at least 50 fields, which is often viewed as prohibitive
and its size and spacing. Remember, good specimen preparation when done manually. Hence, the interest in image analysis.
procedures are a necessary requirement for obtaining good data.

Conclusions
Although all of the most important stereological rules for measuring
matrix microstructures were well known and described in Dr.
Underwood’s 1970 book (and elsewhere), their application was
chiefly limited to research studies. Production metallographers did
not adopt these methods because of the time required to make
measurements manually.

The incredible growth in microcomputer power and in video

technology over the past fifteen years has permitted development
of very capable, reasonably priced user-friendly image analyzers,
such as BUEHLER’s OMNIMET® Advantage System. Image analysis
permits collection of microstructural data (some of which are difficult
to obtain manually) rapidly, painlessly, and with sufficient sampling
to provide good statistical data. Future copies of Tech-Notes will
be devoted to the subject of image analysis and its applications.

Tech-Tips
Question: We have been making delta ferrite measurements using
the AMS 2301 procedures for many years, starting with the occupied
squares method and then with the point counting methods added
in versions B and C of the method. Why have these methods been
changed, which should I use?
Answer: When AMS 2301 was developed, it contained a single
method for estimating the amount of delta ferrite in a specimen
(many specifications require less than 5% δ), called the “occupied
squares” method. In this method, a grid consisting of 500 small
squares (made by intersecting horizontal and vertical lines) was
placed over the microstructure and a count was made of the number
of “boxes” fully occupied by delta ferrite, the number 3/4 occupied,
1/2 occupied, and 1/4 occupied. The results were tabulated and
added up to determine the fraction of the grid area covered, or
“occupied,” by the delta ferrite. This is basically an areal analysis
method and it gives a rather precise measurement for that particular
field, but only a few fields can be evaluated because it is so tedious.
Measurement precision for the specimen is influenced much more by
the field-to-field variability of the delta ferrite than by the precision
of measuring each field. Thus, it is necessary to measure the delta
on more fields. This method is too tedious for evaluating 20 or
more fields, unless there is virtually no delta present. But, if that is
the case, it may not be necessary to make a measurement at all.
However, when the amount of delta is close to the typical 5% limit,
the occupied squares method is too slow for practical use.

AMS 2301 next added a point counting method using a 500 point
grid, which was later changed to 100 points. The optimum point

BUEHLER Worldwide Headquarters BUEHLER France BUEHLER Japan

41 Waukegan Road [email protected] [email protected]
Lake Bluff, Illinois 60044-1699 USA
P: (847) 295-6500 BUEHLER United Kingdom BUEHLER China
www.buehler.com | [email protected] [email protected] [email protected]
Connect with us: BUEHLER Canada BUEHLER Asia-Pacific
BUEHLER Germany
[email protected] [email protected] [email protected]

© 2015 BUEHLER, a division of Illinois Tool Works Inc. Printed in U.S.A. FN01024_0615