Measurement of Grain Size

George F. Vander Voort

Director, Research & Technology
Buehler Ltd
Lake Bluff, Illinois USA
5/16 Sieve
Actual Grain Shapes 4 Sieve

3.3 mm
3.3 mm

Separation of
grains by sieving
after liquid metal
embrittlement of 3.3 mm
3.3 mm Brass in Hg 14 Sieve

8 Sieve
 Actual Grain Shapes

1.1 mm 1.1 mm

SEM secondary electron images of individual Brass

grains separated by liquid metal embrittlement with Hg.
 Types of Grain Sizes

• Non-twinned(ferrite, BCC metals, Al)

• Twinned FCC Metals (austenite)
• Prior-Austenite
(Parent Phase in Q&T Steels)
• Number of Grains/inch2 at 100X: G
• Number of Grains/mm2 at 1X: NA
• Average Grain Area, µm2 : A
• Average Grain Diameter, µm: d
• Mean Lineal Intercept Length, µm: l
 Comparison Chart Ratings
Shepherd Fracture Grain Size Ratings
Jeffries Planimetric Grain Size
Heyn/Hilliard/Abrams Intercept Grain Size
Snyder-Graff Intercept Grain Size
2D to 3D Grain Size Distribution Methods
 n = 2 G-1

n = number of grains/in2 at 100X

G = ASTM Grain Size Number
G n G n
1 1 6 32
2 2 7 64
3 4 8 128
4 8 9 256
5 16 10 512
 ASTM Standards for Grain Size

ASTM E 112: For equiaxed, single-

phase grain structures

ASTM E 930: For grain structures

with an occasional very large grain
ASTM E 1181: For characterizing
duplex grain structures
ASTM E 1382: For image analysis
measurements of grain size, any type
 Determine for any standard that generates
numerical data the precision and bias by use
of interlaboratory “round robins”.

Accuracy generally cannot be determined as

the true values being measured cannot be
determined by any referee method.
 Look at a properly etched microstructure,
using the same magnification as the chart, and
pick out the chart picture closest in size to the
test specimen. If the grain structure is very
fine, raise the magnification, pick out the
closest chart picture and correct for the
difference in magnification according to:
G = Chart G + Q
Q = 6.64Log10(M/Mb)
where M is the magnification used and
Mb is the chart magnification
n1 = number of grains completely inside the
test circle
n2 = number of grains intercepting the
circle

NA = f[ n1 + (n2/2)]
f = Jeffries multiplier
f = magnification2/circle area
 1
Average Grain Area = A = ——
NA

G = (-3.322LogA) – 2.955
n1 = 68 and n2 = 41
For the preceding micrograph,
n1 = 68 and n2 = 41
And

M2 100 2
f = —— = ——— = 0.497
A 20106.2
 NA = f[n1 + (n2/2)]
NA = (0.497)[68 + (41/2)]
NA = 44.02 mm-2
 1
A = —— = 0.0227 mm2
NA

d = (A)1/2

G = 2.5
 This is an austenitic Mn steel, solution annealed and aged to
precipitate a pearlitic phase on the grain boundaries (at 100X).
There are 43 grains within the circle (n1) and there are 25 grains
intersecting the circle (n2). The test circle’s area is 0.5 mm2 at 1X.
 NA = f[n1 + (n2/2)]

f = [(1002)/5000]

NA = 2[43 + (25/2)] = 111 mm-2

G = [3.22Log10(111)] – 2.954 = 3.8

(Of course, more than one field should be measured to get

good statistical results)
 N = number of grains intercepted
P = number of grain boundary intersections

N
NL = ——
LT

PL = ——P
LT
where LT is the true test line length
 Apply a test line over the microstructure
and count the number of grains intercepted
or the number of grain boundary
intersections (easier for a single-phase grain
structure). After you count N or P, divide
that number by the true line length to get
NL or PL.
 Intercept Counts (N)

1/2 1 1 1/2
1 1 1

The test line intercepted 5 whole grains and the line ends fell
in two grains. These are weighted as ½ an interception. So the
total is 6 intercepts (N=6).
 Intersection Counts (P)

1 1 1 1 1 1

The test line has intersected 6 grain boundaries. The ends

within the grains are not important in intercept counting.
So, P=6 for the intercept count.
 1 = ——
Mean Lineal Intercept, l = — 1
NL P L

G = [6.644Log10(NL or PL)] – 3.288

G = [-6.644Log10(l)] – 3.288

Note: Units are in mm-1 (for NL and PL) or mm (for l)

 If the grain structure is not equiaxed, but
shows some distortion of the grain shape, use
straight test lines at various angles, or simply
horizontal and vertical with respect to the
deformation axis of the specimen.
Alternatively, you can use test circles, such as
the ASTM three-circle grid (three concentric
circles with a line length of 500 mm). This
test pattern averages the anisotropy.
Example of three concentric test circles
for point counting.
To illustrate intercept counting, note that there are 41, 25 and 20 grains
intercepted (N) by the three concentric circles.
 LT = 11.4 mm

N = 41 + 25 + 20 = 86

86 = 7.54 mm-1
NL = ——
11.4
1 = 0.133 mm
l = ——
7.54
G = [-6.644Log10(0.133)] – 3.288 = 2.5
Intercept Grain Size Example – Single Phase

This is a 100X micrograph of 304 stainless steel etched electrolytically with

60% HNO3 (0.6 V dc, 120 s, Pt cathode) to suppress etching of the twin
boundaries. The three circles have a total circumference of 500 mm. A count
of the grain boundary intersections yielded 75 (P=75).
Intercept Grain Size Example – Single Phase

75
PL = ——— = 15 mm-1
500/100

1 = 0.067 mm
l = ——
15

G = [-6.644Log10(0.067)] – 3.288 = 4.5

 Intercept Grain Size Example:
Single Phase Twinned Grain Structure

The 100X micrograph is that of a twinned FCC Ni-base superalloy, X-750, in

the solution annealed and aged condition after etching with Beraha’s reagent
which colored the grains. This is a much more difficult microstructure for
intercept counting. The three circles measure 500 mm and P is 63
(intersections with twin boundaries are ignored).
 Intercept Grain Size Example:
Single Phase Twinned Grain Structure

63
PL = ——— = 12.6 mm-1
500/100

1 = 0.0794 mm
l = ——
12.6

G = [-6.644Log10(0.0794)] – 3.288 = 4
N = Number of grains intercepted

LT = Test line length/Magnification

VV = Volume fraction of the phase

VV (LT)
l = ———
N
 This 500X micrograph of Ti-6242 was alpha/beta forged and alpha/beta annealed,
then etched with Kroll’s reagent. The circumference of the three circles is 500 mm.
Point counting revealed an alpha phase volume fraction of 0.485 (48.5%). 76 alpha
grains were intercepted by the three circles.
 (0.485)(500/500)
l = ———————— = 0.006382 mm
76

G = [-6.644Log10(0.006382)] – 3.288 = 11.3

Because the grain size of hardened high speed tool steels is generally around
G = 9 to 12, Snyder and Graff proposed an alternate intercept method. In
this range NA changes by a factor of 10 and the mean lineal intercept length,
l, varies from 14.1 to 5 µm.
To increase the sensitivity to these small variations, they suggested doing an
intercept count at 1000X using a 5-inch (127-mm) test line. The number of
grains intercepted by the line is counted. This is repeated for 10 random
placements of the test line. The average value of the number of intercepted
grains is the S-G intercept grain size number.
ASTM G can be calculated from the NIS-G value:

G = [6.635Log10(NIS-G)] + 2.66
The 1000X micrograph above of a high speed steel in the quenched and tempered
condition has been etched with 10% nital. Two 5-inch (127-mm) lines have been
drawn and the number of intercepted grains were counted. For each line there
were two tangent hits (each weighted as (1/2). One line had 12 intercepts and the
other 13. So, N was 13 and 14, with an average of 13.5 (NIS-G = 13.5) and G=10.2.
 A number of ASTM E-4 members counted intercepts using the three-
circle grid and then counted the grains within a test circle, and
intersecting the test circle, on seven micrographs. Three were at
different magnifications for a ferritic stainless steel and four were at
different magnifications for another ferritic stainless steel. All images
were taken from the same region. The people did not calculate the grain
size; they only collected the raw data. Prior to that, they used a
comparison chart, plate I of E 112, to estimate the grain size of each
micrograph.
A few people digitized the images and measured the grain size with
image analysis systems.
Examples of the micrographs are shown on the next slide. For the
counting, the micrographs were enlarged to 8 x 10 inches. Random grid
placement was used for the intercept method, but for the planimetric
method, the template contained five test circles, so the placement on the
micrograph was not completely random, but forced.
Examples of the micrographs used for the round robin. There were three
magnifications for the one at left and four for the one at right. Grain
boundary delineation was excellent.
 Distribution of grain size by number % and area % (preferred) for the left
image in the previous slide (image analysis results). There is a slight degree of
duplexity in the distribution.
 Distribution of grain size by number % and area % (preferred) for the right
image in the earlier slide (image analysis results). There is less duplexity in the
distribution than for the other specimen.
Results for the first specimen at three magnifications.
Results for the specimen with four magnifications.
 “wild” value

A plot of the planimetric grain size measurement vs. the intercept grain size
measurements for all specimens reveals a normal scatter around the one-to-
one trend line (except for one point) indicating no bias between the methods.
If the true magnification is not used, but all images are assumed to be at 100X, the
different magnifications and give a wider spread of apparent grain sizes. Note that
the comparison chart ratings are consistently lower than the measured values by
0.5 to 1 G value indicating bias in the comparison chart ratings.
Naturally, when the intercept measurements are plotted vs. the comparison chart
estimates of G (similarly to the last slide where the planimetric data was used), the
same bias in the comparison chart data is observed.
Plot of the relative accuracy for the planimetric measurements indicating that
about 1000 grains must be counted to get <10% RA.
 For the intercept method, <10% RA can be obtained by counting about 400
intercepts or intersections. Counting with the planimetric method is more tedious
as the grains must be marked off to get an accurate count.
 Plot of the %RA as a function of the average count per grid placement
(per field). Counting errors start to results when the count exceeds about
50-60 per field.
 Inadequate etch times do not reveal the grain structure so bias is
created as the grain size appears to be greater than it is. The above
example was for ferrite grains in low-carbon sheet steel.
 It is possible to make measurements of the
diameter, lineal intercept lengths, or areas of
grains and plot these data in histogram
fashion. Many procedures have been
developed to translate these measurements
on the two-dimensional sectioning plane to
develop three-dimensional grain size
information. Nearly all models utilize some
simplifying assumptions about shape, such as
spherical grain shapes.
Grain structure of 304 austenitic stainless steel etched with 60% HNO3 at 0.6
V dc, Pt cathode, 120 s (this does not bring up twin boundaries) used for the
following grain size distribution study.
 A log plot of the intercept length vs. the number percent per class yields a good
representation of the distribution. Note the slight skew of the data ( 1) while the
kurtosis, 2, is close to the ideal value of 3 for a Gaussian distribution.
A linear plot of the data does not reveal a good distribution as it is skewed more
to the right and the kurtosis is higher.
 Three specimens of an experimental 5% Cr hot-work die steel were
analyzed for their grain size distribution. This one was austenitized at 1950
°F (1066 °C). The others were austenitized at 1925 and 1975 °F (1051 and
1079 °C). The specimens were quenched to 1300 °F (704 °C), held 1 h to
precipitate a pearlitic like constituent at the grain boundaries and air
cooled. They were etched with glyceregia.