6.6.

6 Fiber volume (Vf) fraction

6.6.6.1 Introduction

The fiber volume (expressed as a fraction or percent) of cured polymer-matrix composites is

commonly obtained by matrix digestion, ignition loss, areal weight, and image analysis methods.
These methods
generally apply to laminates fabricated from most material forms and processes, but the areal weight
method cannot be used for filament wound material or other forms that do not consist of discrete
individual plies. Each method has its benefits and drawbacks. Other less common methods will not be
discussed.

6.6.6.2 Matrix digestion

The method of matrix digestion is covered under ASTM Test Method D 3171 “Fiber Content of Resin-
Matrix Composites by Matrix Digestion” (Reference 6.6.6.2). The technique is based on digestion of
the matrix by a suitable liquid which does not attack the reinforcing fibers. Depending on the resin,
three different procedures are used: Procedure A, concentrated nitric acid; Procedure B, aqueous
mixture of sulfuric acid and hydrogen peroxide; and Procedure C, a mixture of ethylene glycol and
potassium hydroxide. For example, epoxies generally respond well to all three procedures. Although
toughened systems respond better to procedure B, some fiber types are attacked more by B than by
A. BMI’s, polyimides, and thermoplastics usually respond well to procedure B. Aramid fibers are
attacked by both A and B, and therefore, procedure C works best on aramid fiber composites.

Possible causes of error:

If the fiber is significantly attacked by the digestion fluid, results will be erroneous. It is recommended
to validate the process by testing a control sample of only fiber to determine mass change of fiber
during test.

Some toughened resin systems have additives such as elastomers or thermoplastics. If these
additives are not dissolved by the digestion fluid, they may cling to the fiber causing erroneous
results.

Incomplete digestion of the resin.

Sample size must be large enough to be representative and weighed accurately.

Accuracy is dependent on accuracy of density measurements.

6.6.6.3 Ignition loss

The method of ignition loss is described in standard test method ASTM D 2584 “Ignition Loss of
Cured Reinforced Resins.” The technique determines ignition loss of cured polymer-matrix
composites which can be considered to be the resin mass. A weighed specimen is heated until the
resin matrix is oxidized and converted to volatile materials. After removing any remaining ash, the
residue (reinforcing fiber) is weighed and the percent loss is calculated. Fiber density and composite
density (to three significant figures) are required in order to calculate fiber volume.

Possible causes of error:

If the fiber gains or loses weight under the conditions of the test, the results will be erroneous. (For
this reason, this procedure is not appropriate for aramid fibers and requires special temperature
controls for carbon fibers.)

Fillers, if present, must be oxidized with resin.

Incomplete decomposition of resin (and fillers, if present) during the test.

Any volatiles such as water, residual solvent, etc., will cause error unless they are small enough to be
ignored.

If the sample is heated too rapidly, mechanical loss of noncombustible residue (fiber) can occur,
causing erroneous results.

6.6.6.4 Areal weight/thickness

Sections 6.6.6.2 and 6.6.6.3 discussed methods for determining fiber volume by destructively
separating the fibers from the matrix through chemical or thermal means and measuring the mass of
the fibers in a sample. As discussed in Section 2.4.3 on data normalization, there is a relationship
between laminate (or specimen) thickness and fiber volume for given values of fiber areal weight and
fiber density. This fact provides the basis for another method of fiber volume determination, which is
not destructive.

In general the method involves measuring the laminate or specimen thickness and calculating the
fiber volume using this measured thickness, the number of plies in the laminate, and previously
determined values of fiber areal weight and fiber density. Equation 6.6.6.4(a) or an equivalent form is
used.

6.6.6.4(a) where

Vf = fiber volume fraction

FAW = fiber areal weight (mass per area per ply)
n = number of plies in the laminate
t = measured laminate (or specimen) thickness
f= density of the reinforcing fiber
k = units conversion factor (if required)

The calculated fiber volume is the reinforcing fiber’s contribution to the total volume. Although void
content does affect the laminate thickness (and hence fiber volume), it is not a factor in the calculation
since it contributes to total volume in the same way as the resin or any other non-reinforcement
component. Since the calculation requires the number of plies, the method is applicable only to
material forms with distinct plies for which fiber areal weight can be determined.

This general procedure is documented in SRM 10R-94 (Reference 6.6.6.4(a)), which references other
SACMA Recommended Methods for determination of fiber areal weight and fiber density. The
specimen defined by this method is a laminate panel, but the concept could be extended to individual
test specimens or to application parts. The method notes that the fiber areal weight and fiber density
used in the fiber volume calculation must be representative of the sample (panel, specimen, or part)
under evaluation. This is an important point. Although resin content may typically be the major factor
affecting fiber volume, fiber areal weight and fiber density variations can also have a significant effect
on the accuracy of this method. “Typical” or “data sheet” values for these parameters should not be
used in the calculation. SRM 10R-94 recommends that, as a minimum, fiber areal weight of the
individual prepreg roll and lot average fiber density be used. If careful thickness measurements are
taken (see Section 6.6.5) and appropriate fiber areal weight and fiber density values are used, this
method can be quite accurate, and potentially more reliable than destructive methods that depend
heavily on laboratory technique.
For certain types of specimens the accuracy of thickness measurements may not be sufficient to yield
accurate fiber volumes. In particular, very thin specimens present the problem of obtaining three
significant figure precision. In addition, thickness cannot be reliably measured for specimens with
irregular surfaces. As an alternate to measuring specimen thickness directly, it may be calculated
using equation 6.6.6.4(b):

6.6.6.4(b) where
tc = calculated laminate (or specimen) average thickness
M = mass of the specimen
A = surface area of the specimen
c = measured density of the composite specimen

Combining equations 6.6.6.4(a) and 6.6.6.4(b) yields the following expression for fiber volume fraction
interms of laminate specimen mass, density, and area:

6.6.6.4(c)

When mass is in grams, FAW in g/m 2, densities in g/cm 3, and area in in.2, then k = 1/1550.

For this variation of the method, the specimen must be extracted from the laminate or part such that
its edges are cut cleanly without any fraying. Ideally the specimen should be rectangular to permit
area Calculation from specimen length and width. If possible, the specimen should be cut at least one
inch from any laminate (part) edge to minimize edge effects caused by fiber wash. The specimen
mass is obtained by weighing on a balance with adequate precision, and density is measured using
archemedian or other suitable methods (see Section 6.6.4 of this volume). More detail may be found
in Reference 6.6.6.4(b).

Possible causes of error:

Inaccuracies in measured thickness caused by use of inappropriate measuring equipment,

particularly on textured surfaces (see Section 6.4.2).

Inaccuracies in calculated surface area (for the area variation of this method) caused by edge
damage induced during specimen cutting or by inappropriate measuring equipment or technique,
especially For irregularly shaped specimens.

Wide variation in measured thickness across the specimen (in which case the calculated average
fiber volume is not representative of all areas of the specimen).

Significant fiber wash (spreading) during cure (which causes an effective reduction in fiber areal
weight in the specimen compared to the value measured for the prepreg).

Use of fiber areal weight and/or fiber density values that are not representative of the fiber in the
specimen.
6.6.6.5 Determination of fiber volume using image analysis

6.6.6.5.1 Background

The method of image analysis offers a technique of measuring fiber volume that eliminates the
creation of waste chemicals while providing information about void volume laminate orientation and
through thickness fiber distribution. The basic assumption for this technique is that the evaluation of
the two dimensional distribution of fibers through a random cross section is representative of the
volumetric fiber distribution. This assumption is valid for fibers of constant cross section, such as is
found in tape laminates, but is not valid for woven laminates. This technique works well for carbon
fibers in a polymer matrix, and for other fiber/matrix combinations where adequate contrast can be
achieved. It does not, for example, work as well for glass fibers, since the low contrast between the
glass fibers and surrounding matrix makes accurate measurements very difficult. There are no
industry standard test methods for this type of evaluation. Therefore, this section will describe the
procedure in general terms. Computer software for image analysis is commercially available.

6.6.6.5.2 Apparatus

This technique requires the use of metallographic specimen preparation equipment, a reflected light
microscope with a magnification of at least 400 times which has the capability of porting the image to
a digital camera, a computer with image acquisition card and image analysis software. While
automated image acquisition systems are available, this analysis can also be performed by manual
specimen translation and focusing. The use of software macros can reduce the time required to
process a fiber volume measurement. Macros allow the user to automate repetitive software
instructions.

6.6.6.5.3 Specimen preparation

A small section of laminate is prepared using standard metallographic techniques. A typical specimen
is ¾ inch (2 centimeters) on a side. The cross section is taken across the thickness of the laminate.
The orientation of the cross section with respect to the laminate is dependent on the orientation of the
fibers in the laminate being evaluated. The image analysis technique can be used to evaluate fiber
volume in laminates with ply orientations from 0 to approximately 60. At angles above 60 degrees the
fiber edge definition becomes distorted by the subsurface fibers. If a laminate with 0/90 lay-up is
evaluated, the laminate can be sectioned at 45to enable evaluation of the fiber volume in all plies. If
fiber volume is to be measured on a 0/60 laminate, the cross section can be oriented at 0so that all
plies can be evaluated.

The surface of the polished specimen should display a clear delineation between the fibers and the
matrix. The fiber volume measurement should be made at as high a magnification as possible. This to
some extent depends on the fibers being examined, but for most fibers a magnification of at least 400
to 1000 times should be used. This should get 30 to 100 fibers in the field of view. As these are areal
measurements being performed, the partial fibers can be included in the analysis. This would not be
true if the individual fiber area were being determined. A typical image is shown in Figure 6.6.6.5.3.
6.6.6.5.4 Image analysis

The objective of the fiber volume image analysis technique is to discriminate between the fibers and
matrix. The image is acquired as a gray scale, threshold intensity is used to select the delineation
point between the fibers and matrix. The threshold level can be determined by evaluating a histogram
of the image as shown in Figure 6.6.6.5.4(a). Typically once the threshold is selected for a given
cross section it does not have to be altered as additional images are acquired for that specimen. It is
good practice to display the histogram while acquiring images to confirm the veracity of the threshold
level. The threshold level is used to convert the grayscale image to a binary image (Figure
6.6.6.5.4(b)) where the fibers will be either black or white and the matrix will be the opposite (white or
black). The computer then counts the number of black and white pixels, and the ratio of the fiber
pixels to the total number of pixels in the image is the fiber volume.

While automated image systems can be programmed to analyze the entire cross section, this may
require as many as 1000 images. Accurate results can be obtained from manually operated systems
using from 20 to 50 samples. Testing has shown that where the fibers/resin are evenly distributed, the
mean fiber volume converges to a constant value in as few as 20 samples. The manual sampling
should be distributed throughout the cross section.

The typical steps in the analysis of a single frame are:

1. Position the specimen. (Manually/automatically move the stage on which the specimen rests.)
2. Focus the microscope. (Manual focusing requires the use of a real time monitor on older
computers. Newer computers have acquisition rates fast enough to preclude the use of a monitor.)
3. Acquire an image. (The image used to measure the fiber volume may be a single frame or an
average of multiple frames. Integrating several images can offset low luminescence of an image.
Examination of the histogram will indicate if the image is suitable for evaluation.)
4. Identify pixels corresponding to fibers. (The histogram should be checked to insure that the correct
threshold value has been selected.)
5. Create a binary image. (The pixels with values greater than the threshold value will be black and
the pixels with values less than the threshold value will be white.)
6. Count the number of white and black pixels. (Typically it will be necessary to obtain a count of only
one of the colors. The total number of pixels in the frame remains constant so the number of fiber
pixels for a given image is all that needs to be recorded.)
The time required to process a fiber volume measurement can be accelerated by the use of macros
within the image analysis program. After the microscope has been focused (step 2), a macro can be
initiated via a single keystroke which initiates the acquisition sequence (steps 3 through 6). The
histogram can be displayed so that the operator can verify the adequacy of the threshold value
selected. Some imaging software programs will have an automatic threshold operation.

6.6.6.5.5 Sources of error

Out of focus or dirty optics can distort the image which in turn will give inaccurate results.

Poor metallographic preparation techniques of the cross-section surface makes accurate

thresholding difficult.

Insufficient magnification will result in poor definition of the fibers.

Poor microscope lighting or incorrect use of blank field image will distort the intensity distribution.
The filament in the incandescent bulb used to illuminate the surface may not be uniform. This would
result in a non-uniform distribution of light on the specimen surface and yield a distorted histogram.
This can be corrected by creating a blank field image that is subtracted from the acquired image. The
blank field can be taken either with the microscope slightly out of focus or in focus on a clean area of
the polished mounting material surface. Figures 6.6.6.5.5(a) and (b) illustrate this correction.