Bai, X., Smart, P.

& Lag,

X. (1994). Gdotechnique

Polarizing

44, No. 4, 175-180

micro-photometric

X. BAI*, P. SMART*

analysis

and X. LENG*

La comhinaison de la rotation synchrone des polariseurs dun microscope optique et de Ianalyse

dimage digitalis&e permet dohtenir la distribution
des orientations et des cartes detaillees des directions dorientation.

The combination of synchronous rotation of the

polars in an optical microscope and digital image
analysis provides orientation distributions and
detailed maps of the directions of orientation in the
micrographs.
KEYWORDS:
anisotropy; clays; fabric/structure of
soils; microscopy.

POLARIZING MICRO-PHOTOMETRIC
ANALYSIS (POLMAP)
The transfer function of the intensity of light
through a polarizer, a thin section of a birefringent material, and an analyser is

INTRODUCTION
Optical microscopy
has been widely used in the
investigation
of soil structure
in geotechnical
engineering since Mitchell (1956) first applied it to
this field. Much attention has been paid to the
analysis of orientation,
because certain kinds of
orientation data, if properly used, can be valuable
in the elucidation of changes of soil structure in
different
mechanical
states.
However,
when
viewed between crossed polars, the matrix of clay
soil often shows diffuse light and dark patches
which represent information
that has been difficult to extract by conventional
microscopy.
By
using a special type of optical microscope
and
computer
analysis system, new methods
have
been developed to extract more of this information.
In particular, the light and dark patches can
now be mapped on the basis of local preferred
orientation
within the patches,
histograms
of
orientation distributions
can be obtained, and the
overall orientation
of the field of view can be
measured.
A complication
might arise if the
sample contained
assemblages
of clay particles
that were large enough to be resolved and that
had different effective optical properties in three
different directions; this might restrict the universal applicability
of the methods. However, these
methods
have already proved valuable in the
analysis of microstructural
changes during three
studies of the deformation
of kaolin (Bai, 1992;
Smart & Tovey, 1992). This Paper explains these
micro-photometric
polarizing
methods
of
analysis.

I = 0.5 sin2 20 sin2 612

(1)

where 0 is the angle between the section axes (i.e.

the axes of birefringence)
and the axes of the
polars, and 6 is the phase difference between the
ordinary and extraordinary
rays after the light
has passed through the thin section. If a thin
section is viewed between crossed polars, it is
known that extinction
occurs only when the
section axes are parallel to or perpendicular
to
the axes of the polars. When a quarter-wave plate
is used in such a position that the angle between
the axis of the quarter-wave
plate and the axis of
the polarizer is 45, the transfer function becomes
I = 0.5 (1 - sin 28 sin 6)

(2)

In this case, full extinction occurs only when both

the section axes are at 45 or 135 to the axis of
the polarizer and 6 is sin 1 1 or sin- 1 ( - 1). The
second of these conditions is rarely satisfied, but
the partial darkening predicted by equation (2) is
sufficiently pronounced to be of value here.
The methods reported here use partial darkening effects such as that discussed above by
comparing several micrographs
of the same area
digitized under different polarizing conditions. In
most optical microscopes, changing the polarizing
conditions involves rotating the thin section, but
it is virtually impossible to adjust the centring of
the axis of rotation of the thin section with suflicient accuracy, so problems of registration
arise.
To avoid these, in the present work, the thin
section is held stationary, and different polarizing
conditions
are achieved by adjusting the polarizing system itself.

Manuscript
received 29 September 1993; revised manuscript accepted 17 June 1993.
Discussion on this Paper closes 1 July 1994; for further
details see p. ii.
_
* University of Glasgow.

175

176

BAI, SMART

A Leitz Dialux-pol
microscope
was used, in
which the polarizer, the analyser and a slot for
a compensator
are all linked by an arm called
the synchronizer,
so that they can be rotated
synchronously.
A Prior-Swift
MPB124
Polar
Microscope was also used, in which similar synchronous
rotation
was achieved
by internal
gearing.
Figures l(a) and l(b) shows a thin section of
sheared kaolin examined between crossed polars.
A failure plane is shown because its inclusion may
help to explain the method. In Fig. l(a) the synchronizer is set at 45, and the polars are crossed
on the diagonals,
so that the bright areas are
areas in which the clay particles are aligned on
the square, i.e. either horizontally
or vertically
with respect to the field of view. (These micrographs are of a vertical section that has been
turned through 90.) In Fig. l(b) the synchronizer
is set at 90, and the polars are crossed parallel to
the sides of the micrograph,
so the bright areas
are areas in which the clay particles are aligned
on the diagonal. In the first stage of analysis,
these two micrographs
are digitized, the grey
levels (i.e. intensities of illumination)
being measured at every pixel on a 512 x 512 grid on a
O-255 grey scale using a Micro-Semper
image
analysis sytem. The digitized area used here was
0.47 mm square, giving a spacing between pixel
centres of 0.92 pm. These two digitized micrographs are denoted xl and x2 respectively. Their
comparison
distinguishes
between areas oriented
on the square and on the diagonal. Each pixel is
coded as 1 if it is brighter in xl than in x2, on the
square, 2 if it is brighter in x2 than in xl, on the
diagonal, and 0 otherwise, i.e. undecided.
If the digitization is noisy, it might be necessary
to code pixels as undecided if the grey levels in xl
and x2 are approximately
equal. Alternative techniques are to integrate
the digitization
over
several scans of the image, or to filter out the
noise. However, none of the elaborations
have
seemed necessary in the work done so far.
After this preliminary
coding, four more digitized micrographs
are used to distinguish
horizontal
from
vertical
and
down-left
from
down-right.
All these four micrographs
are taken
with a quarter-wave
plate inserted in the compensator slot and rotating synchronously
with the
polars.
In the second stage of analysis, only pixels
coded 1 are considered,
and two micrographs
taken with the synchronizer
set at 135 and 45
are used (Figs l(c) and l(e)). In one of these
micrographs
the brighter areas are horizontal;
in
the other they are vertical: a practical method of
deciding which is which is to experiment
with
failure planes in which the particle alignment can
be assumed safely. These two micrographs
are

AND LENG

denoted ql and q3 respectively. Each pixel that

was originally coded 1 is then recoded as 1 if it is
brighter in ql than in q3, i.e. horizontal, 3 if it is
brighter in q3 than in ql, i.e. vertical, and 0 otherwise, i.e. undecided.
In the final stage of analysis, even pixel coded 2
is recorded using two micrographs
with the synchronizer set at 90 and 0 (Fig. l(d) and l(f)), and
labelled such that the new coding is 2 if it is
brighter in q2 than in q4, i.e. down-left, 4 if it is
brighter in q4 than in q2, i.e. down-right,
and 0
otherwise,
i.e. undecided.
For the analytical
method used here, monochromatic
light with a
wavelength matched to that of the quarter-wave
plate should in theory be used, but white light
seems to be satisfactory in practice.
During analysis, a convenient method of presenting the results is to colour the horizontal
pixel 1 green, the down-left pixel 2 yellow, the
vertical pixel 3 red, the down-right
pixel 4 blue,
and undecided 0 black. The colours were chosen
because thin sections such as the one used here
frequently shown yellow and blue for the pixels
coded 2 and 4 respectively when viewed in white
light with the synchronizer
set at 90 and using a
full-wave compensator.
Under those conditions,
the other areas tend to be pink, leading to green
vertical and red horizontal.
Alternatively,
the
mapped directions can be ruled. Fig. 2 shows the
result of the analysis, where the down-left
and
down-right
areas have been ruled with inclined
lines, the other areas having been left blank.
The overall anisotropy of the field of view can
be assessed by considering
the number of pixels
mapped into each direction.
For the field discussed the result was undecided 525, horizontal
237309, down-left 2193, vertical 348, and downright 21769.
Following
Smart (1966), an anisotropy
index
A2 obtained by using POLMAP
can be defined
as
A2=l(H+

V)-(R+QI

H+V+R+L

(3)

where H, V, R and L are the number of pixels

mapped into the directions
horizontal,
vertical,
down-left
and down-right
respectively.
In this
above example, approximate estimates are

R+L=23962
H+V=237657
which leads to
A2 = 0.81
In addition
to this overall measurement
of
orientation,
POLMAP
gives extra information,
such as sizes of patches, which may be very useful

BAI, SMART AND LENG

178

POLMAP4

----

POLMAPB

Fig. 3. Polar histograms for Fig. 1

Fig. 2. Field of Fig. I mapped using POLMAP

in the analysis of images. As mentioned above, a
result of analysis was obtained
in which each
pixel was coded as 1, 2, 3, 4 or 0. The number of
pixels for every patch can then be counted at
every line, so an average of the numbers of pixels
for the patches in the horizontal direction, which
is defined as horizontal chord size, can be calculated. In the same way, vertical chord size can be
obtained. For this example, the chord sizes are 56
for the horizontal and 62 for the vertical.

ADVANCED POLARIZING MICROPHOTOMETRIC MAPPINGS

When a parallel beam of monochromatic
light
passes through
a polarizer,
a thin section, a
quarter-wave
plate, and an analyser which is
crossed with the polarizer, the transfer function I
of the intensity of the light is given by equation
(2) where 8 now refers to the inclination of one of
the section axes of a pixel. Suppose the synchronizer is set in turn to O, 45, 90 and 135: the
intensity of the light at each setting can then be
calculated from
I, = 0.5 (1 - sin 28 sin 6) = q2

and 4(f) the direction of preferred orientation

at
each pixel can be obtained. For example, Fig. 3
shows the polar histogram
of angles from the
same area as was used when POLMAP
was discussed, which referred to the true horizontal and
vertical. This new method, called POLMAP4,
gives the direction of maximum preferred orientation to be 14 from the true horizontal.
If a map is required in addition to a complete
distribution
of orientation,
the degrees
are
divided into groups, which in Fig. 4 are horizontal (from 0 to 22.5 and from 157.5 to 180),
vertical (from 67.5 to 112.5), down-left
from
22.5 to 67.5) and down-right
(from 112.5 to
157.5) respectively, each group being given the
same colour as for POLMAP.
The result is
shown in Fig. 4 (for the same reason as in Fig. 2,
the down-left
and down-right
areas have been
ruled with inclined lines). Comparison
of Figs 2
and 4 gives very satisfactory results. If desired, the
degrees could be divided into a larger number of
groups (say 8 or 12) for a finer mapping.
Taking the analysis further, a new index consistency ratio (CR) (see Curray, 1956) is defined as
CR = &SX + S,)/S
fJ = 0.5 tan-

(5a)

(S,/S,)

where the orientation

(5b)
at each pixel is represented

(4a)

I,,

= 0.5 (1 - cos 28 sin S) = q3

(4b)

I,,

= 0.5 (1 + sin 28 sin 6) = q4

(4c)

I ia

= 0.5 (1 + cos 20 sin 6) = ql

(4d)

Therefore
tan 28 =
sin

6 =

I - 10
9o
I 135 - I,,

JCU90

1,)

(4e)
+

(1135

I,,Yl

Pf)

After all of the four micrographs ql, q2, q3 and

q4, have been digitized, by using equations 4(e)

Fig. 4. Field of Fig. 1 mapped using POLMAP4

POLARIZING

MICRO-PHOTOMETRIC

by a unit vector, S, is the sum of the components

of these vectors on the horizontal
axis at every
decided pixel after the angle is doubled (i.e. S, =
cos 28, + cos 28, + . . .); S, is the sum of the
components
on the vertical axis at every decided
pixel after the angle is doubled (i.e. S, = sin 28,
+ sin 20, + ..); and S is the sum of the unit
vectors
for all decided
pixels. (This is the
unweighted
version of the CR, which is also
known as the mean resultant vector.) The CR
reflects the degree of preferred orientation of particles in the viewed area. In this example, CR is
0.92, a little higher than A2.
POLMAP3
comes from the same theory as
POLMAP4,
and produces the same results from
only three micrographs.
According to equation
(2), when the synchronizer
is set at O, 120 and
240, the intensities of the light passing through
the thin section are determined by
I, = 0.5 (1 - sin 28 sin 6)

(6a)

I 120 -- 0.25 [2 + (sin 20 + J3 cos 20) sin S]

(6b)
I 240 = 0.25 [2 + (sin 28 - J3 cos 28) sin S]

A2 (POLMAP)

Fig. 5. CR plotted against A2

suring the degree of anisotropy,

although A2 is
more convenient in some cases.
Figure 6 shows the relation of the A2, calculated from POLMAP,
and the older anisotropy
found
from
polarizing
microindex
Al,
photometric
analysis as

tg 28 =

Al =

Therefore
I 120 + I,,,
J3(Ii,0

-210

(6d)

- I,,,)

sins = *[(IiZO + I240 - 21,)*

+ JCJ3(Ii2,

124~)12

(64

Fig. 3 also shows the polar histogram of angles

from POLMAP3
for the same area as above,
which agrees very well with the result from
POLMAP4. As before, the CR can also be calculated from POLMAP3.
POLMAP3
requires three micrographs
(qO,
qiZO and q&
for the Prior-Swift
MPB124 polar
microscope.
For the Leitz microscope,
three
micrographs
(qO, qlzo and qco) may be used,
because qZbo is unavailable and I,,, = I,,

179

0.6 -

(6~)
respectively.

ANALYSIS

max - min
max + min

where max and min are the maximum

and
minimum intensities of light transmitted
through
the section between crossed polars as it is rotated
(e.g. Morgenstern
& Tchalenko, 1967; Wu, 1958;
Smart, 1966). The photometer
used measured a
circular field of diameter I.5 mm. The values are
higher for A2 than for Al; this is believed to be
because the areas are smaller for A2 than for Al.
1.

0.6 -

0.6 -

5
DISCUSSION

The anisotropy index A2 is compared with the

CR in Fig. 5, which shows the relationship
of
these parameters for a series of normally consolidated undrained shear tests. The CR is more sensitive to the degree of anisotropy than A2, and is
always larger. These analyses
show that A2
depends not only on the strength of anisotropy
but also on the type of anisotropy. Therefore, the
CR seems a more reasonable parameter for mea-

0.4 -

0.2 -

A2 (POLMAP)

Fig. 6. Anisotropy indices: Al from micro-photometer

plotted against A2 from POLMAP

180

BAI,

SMART

CONCLUSIONS

The
new
methods
of polarizing
microphotometric
analysis presented
here are quick
and easy to perform. These methods extend the
older micro-photometric
analysis to improve the
measurement
of overall preferred orientation, and
to provide fuller information
on the orientation
distribution,
with the sizes and shapes of the
mapped patches. They have already proved useful
in experimental work.
REFERENCES
Bai, X. (1992). Microstuctural deformation of clay. PhD
thesis, University of Glasgow.
Curray, J. R. (1956). The analysis of two-dimensional
orientation data. J. Geol. 64, 117-131.

AND

LENG

Mitchell, J. K. (1956). The fabric of natural clays and its

relation to the engineering
properties. Proc. Highw.
Res. Bd 35,693-713.
Morgenstern,
N. R. & Tchalenko,
J. S. (1967). The
optical determination
of preferred
orientation
in
clays and its application
to the study of microstructure in consolidated
kaolin. Proc. R. Sot. A300, 218250.
Smart, P. (1966). Optical microscopy
and soil structure.
Nature 210, 1400.
Smart, P. & Tovey, N. K. (1992). Microfabric
of deformation of soils. Proc. Air Force Ofice of Scientific
Research Contractors Meeting, Particulate Mechanics Subarea, Albuquerque,
16-19.
Wu, T. H. (1958). Geotechnical
properties of glacial lake
clays. Proc. Am. Sot. Cio. Engrs 84, SM3 1732-11732-34.