WO2008033789A2

WO2008033789A2 - Method of measuring fiber length for long fiber reinforced thermoplastic composites

Info

Publication number: WO2008033789A2
Application number: PCT/US2007/078090
Authority: WO
Inventors: Jacqueline Marie Ayotte; Dieter Rolf Bund
Original assignee: Ticona Llc
Priority date: 2006-09-11
Filing date: 2007-09-11
Publication date: 2008-03-20
Also published as: WO2008033789A3

Abstract

A method of measuring fiber length distribution in a long fiber reinforced thermoplastic composite includes liberating the fiber from the composite by thermally decomposing the polymer followed by optical analysis capable of resolving fiber clusters and accounting for curved fibers, in particular using the Hough transform.

Description

METHOD OF MEASURING FIBER LENGTH FOR LONG FIBER REINFORCED THERMOPLASTIC COMPOSITES

Technical Field

The present invention relates to measuring the length of reinforcing fibers in long fiber reinforced thermoplastic (LFRT) materials using automated image analysis. The subject matter of this application relates to that disclosed in Untied States Provisional Patent Application Serial No. 60/823,527 entitled "Process for Forming Improved Fiber Reinforced Composites and Composites Therefrom" of Kenney et al. filed On August 25, 206. The disclosure of Untied States Provisional Patent Application Serial No. 60/823,527 is incorporated herein by reference in its entirety.

Background

LFRT materials are fast becoming the alternative to steel in many applications. Generally, shaped parts are prepared from pellets of coated, long fiber reinforced composite structures by injection molding. The composites are perhaps most preferably prepared by a pultrusion process. United States Patent No. 6,090,319 to Sharma et al, entitled "Coated Long Fiber Reinforcing

Composite Structure and Process of Preparation Thereof describes generally a production process wherein continuous fibers are coated in a die and then cut to the desired length. Suitable coating or matrix polymers include polyolefins such as polypropylene or polyethylene or blends of polyolefins with polyamides and the like. Stabilizers, pigments and so forth are added depending on the desired properties; details are seen in the following patents: United States Patent No. 6,844,059 to Bernd et al. for " Long-Fiber-Reinforced Polyolefin Structure, Process For Its Production, And Moldings Produced Therefrom"; United States Patent No. 6,794,032 to Borgner et al for "Long-Fiber Reinforced Polyolefin Plastic Structure And Shaped Bodies Formed Therefrom"; and also United States Patent No. 6,482,515 also to Berndt et al. for "Colored Long-Fiber-Reinforced Polyolefin Structure And Shaped Articles Produced Therefrom". The pellets, or rod shaped composites generally have a length from 3 to 100 mm, preferably from 4 to 50 mm, and particularly preferably from about 7 to about 15 mm. The diameter of the rod-shaped structure or pellet, is from 1 to 10 mm, preferably from 2 to 8 mm, and particularly preferably from 3 to 6 mm. The fibers suitably have a diameter of from 5-100 microns; typically, around 30 microns or so.

When making the finished part, it is oftentimes desirable, or even critical, to measure fiber length; however, due to relatively complex processing during molding, length degradation always occurs. Fiber length reduction is difficult to quantify and one of skill in the art will appreciate that, absent such quantification, it is difficult to improve existing processes. Indeed, understanding the fiber length distribution to optimize formulation and processing conditions is essential. Currently practiced methods for measuring fiber lengths in the industry include labor intensive methods of measuring from optical micrographs or measuring directly from the fibers. Another method involves stacking sieves, filling them with water and introducing fibers while manually stirring. When stirring is halted, the fibers are assumed to stay horizontal until contacting the sieves, and becoming separated by length. After each sample, the apparatus is dismantled and the fibers are dried and weighed. These methods are time consuming and labor intensive. With the advent of high precision motorized stages, high resolution cameras which perform automated imaging, and new image analysis software capabilities, a procedure to automate fiber length measurement accurately and with repeatability requiring less manual intervention was developed after finding that existing automated optical systems did not adequately account for fiber clusters or curved fibers seen in LFRT materials.

Summary of the Invention

There is provided in accordance with various aspects of the invention, a procedure for determining the length in LFRT materials including: thermally degrading the matrix polymer from a part by application of heat in order to liberate the fiber from the polymer matrix; randomizing the fiber in a liquid medium; filtering the fibers from the liquid medium and dispersing them on a substantially transparent surface; imaging the dispersion of fibers without operator bias and analyzing the images to determine fiber length. The imaging/analysis system is capable resolving fiber clusters through adjustment of settings.

Volume (length) weighted average lengths better represent the fiber length distribution than the number mean for LFRT materials. It has also been found that after about 3000 fibers are measured by the procedure described herein that the average length measured and standard deviation cease to change substantially.

A unique methodology involves color coding touching and overlaying fibers of fiber clusters for classification.

It has been discovered that readily available Hough Transform software can be utilized to analyze fiber dispersions when used with the FASEP plug-in and provide reliable characterization for fiber clusters seen in LFRT-derived samples. It may be preferred in some cases to additionally account for fiber curvature.

Further details and advantages will become apparent from the discussion which follows.

Brief Description of Drawings

The invention is described in detail below in connection with the drawings wherein:

Figures l(a) and l(b) are SEM of glass fiber showing that sizing still exists on the fiber surface after recovery from a molded part;

Figure 2 is a photograph of an automated imaging apparatus and image analyzer;

Figure 3 is an imaged area of dispersed fiber;

Figure 4 is a color coded image indicating classification of dispersed fibers;

Figures S and 6 are color coded images of fiber clusters; and

Figures 7 and 8 are graphical representations of fiber length distributions.

Detailed Description of the Invention

The invention is described in detail below for purposes of illustration without intending to limit the present invention in any way; the spirit and scope of which is set forth in the appended claims.

As noted earlier, the first challenge with quantifying fiber length in LFRT materials has been the recovery of fibers from molded parts without breakage and embrittlement. A fiber recovery process has been determined that yields pliable Fibers some of which are the entire length of the original pellets. This observation indicates high probability that the fibers sustained the recovery method intact. The images in Figure 1 show the fibers before and after 10 hours in a muffle furnace at the noted temperatures. These particular fibers shown were not embedded in a polymer matrix. Sizing is still observed on the fiber surface. At 45O⁰C, the fibers are found to retain their pliability demonstrated prior to heat treatment. However, at 650⁰C the fibers are not as pliable, and at 725⁰C the fibers break easily when maneuvered.

During removal of the sample from the molded part, some fibers are cut. The sample preparation method used in the invention attempts to separate the fibers that have not been cut from those that have. The cut fibers are then discarded. The remaining sample is evaluated for entangled fibers. If entangled fibers are present, they are gently untangled using wire probes. The fibers are then randomized and placed in a Petri dish for image analysis.

A specific procedure for sample preparation is as follows:

(1) Cut a 1" inch square from the area of interest;

(2) Place sample in crucible and ash using a muffle furnace at 45O⁰C over-night (this temperature does not embrittle the fibers);

(3) Use an anti-static device to ensure that the glass sample dish (Pyrex glass Petri dish lOOmmxl Smm - top) is not statically charged. Place the glass sample dish over the crucible and invert crucible so that the ash is in the sample dish but retains its shape;

(4) With a brush, probe, or tweezers, gently part the outer fibers away from the center of the ash to separate those fibers that may have been cut from those which have not. The remaining sample should be - ³/," by %";

(5) Place the crucible over the separated center portion of uncut fibers and invert the glass sample dish over an open plastic bag, which will catch the unwanted saw-cut fibers. Tap the dish to dislodge the fibers that may adhere due to static electricity;

(6) Examine the ash that is left. If large clumps exist, gently separate the clump using probe tips or narrow tweezer tip;

(7) Assemble a vacuum filtration apparatus using a vacuum flask ~

1500ml, a Beuchner funnel with fixed perforated plate (diam.

60mm) and coarse/fast flow filter paper suitable for vacuum filtration;

(8) Place approximately 500ml of water into a beaker (usually a 1000ml beaker is used) and add ~40ml of glycerin. Stir to mix;

(9) Add glass fibers to the beaker and place beaker in ultrasonic bath;

(10) Place distilled water into the ultrasonic bath so that the beaker sits lightly in the bath;

(11) Turn on ultrasonic bath and wait 30 seconds, most of the fibers will separate;

(12) While waiting, wet the filter paper with water and 'seal¹ it against the funnel by creating a slight vacuum. Make sure all holes are covered by the filer paper;

( 13) Place pipet, with an 1 1 mm diameter opening in the beaker and plunge the solution in and out of the pipet ~ 15 to 20 times until all the fibers are suspended and randomized - there should be no clumps. (A fiber-optic light, or other bright light source, may be used to shine into the beaker and illuminate the fibers to confirm suspension and randomization). Note - ash from black parts should be rinsed and filtered first using a similar procedure so that they may be observed randomizing in the solution;

(14) If any clumps are seen, they should be removed from the solution and placed in a separate beaker or dish. Add some solution and gently separate the clump using a probe tip or narrow tweezer tip.

Return to beaker;

(15) Expel any solution from the pipet and move the pipet so that the opening is in the center of the volume of solution. Draw solution in (~20ml):

(16) Increase vacuum in filtration system and bring the pipet over the funnel. Expel the solution in a circular motion in an effort to spread the fibers uniformly over the filter paper;

(17) Rinse the fibers using methanol in a squeeze bottle. Start from the walls of the Beuchner funnel and in a circular motion work to the center of the filter paper. The rinse will dissolve the glycol relatively quickly - a relatively small amount is needed;

(18) The vacuum should be high enough so that the filter paper is observed drying soon after the methanol rinse. The filter paper should be thoroughly dry in 1 minute or less;

(19) Turn off the vacuum and take the funnel off of the flask keeping it upright;

(20) Place a clean and static treated Petri dish over the funnel and quickly invert so that the filter paper falls into the Petri dish. Remove the funnel keeping the filter paper from sliding around the dish (this action will cause the fibers to clump);

(21) Place the Petri dish on a flat surface and the push down on one edge of the filter paper to secure it from moving. Grab the opposite side with tweezers or your finger tips and lift the filter up slightly, keeping it parallel to the bottom surface of the Petri dish. Pull the paper taut, and then bring the opposite sides of the paper together

so that the filter paper folds similar to a book, and snap taut to dislodge fibers from the filter paper. Move grasp on the filter paper 45 degrees and repeat;

(22) If fibers adhere to the filter paper, a soft brush (camel hair) is used to brush them into the Petri dish;

(23) The sample dish should contain randomly aligned fibers and virtually no clumps. If there are clumps: solution may have been expelled from the pipet too fast, too much solution may have been expelled from the pipet, the pipet was maneuvered too slowly over the filter paper as solution was expelled, the dry filter paper slid in the Petri dish prior to it being lifted and 'snapped', or the fiber solution is not dilute enough (add water);

(24) If there are too many fibers in the dish, select a second dish and invert the first one over the second one to reduce the number of fibers (analyze both dishes). The dispersion of fibers should be similar to appended Figure 3, below;

(25) Repeat until at least 3000 fibers are imaged. Add water and glycerin to the beaker if necessary to make up for the solution lost.

A preferred apparatus employed is shown in Figure 2 which includes a

Prior HlOl motorized stage: 4"x 3" travel, repeatedly + lμm, with controller, joystick and holder, as well as a Qlcam monochromatic digital firewire camera: 1392 x 1040 pixels, 4.65μm x 4.65μm pixel size, 1/2" optical format Electronic Shutter, 12-bit, External trigger, Zoom 70XL module with detents/iris. There is further provided an MND44020 Nikon focus Mount and MSS modular support stand, a 15OW halogen transmitted light source with backlight, ImagePro Plus ver 6.0 software, Scope Pro plug-in module, Imaging computer: Windows XP Pro, Pentium 4 3.6GHz processor provided with MS Office 2003 Basic and Pyrex glass Petri dish 100mm X 15mm (top only).

Preferred software to use in connection with the apparatus of Figure 1 is

FASEP Version 1.51 Plug-In for ImagePro Plus, May, 2006, available from IDM Systems, Darmstadt, Germany. This system may be used to analyze clusters with overlaying fibers as well as curved fibers using Hough Transform analysis. For example, the Hough Transform may be used to compute the edge orientation histogram. The Hough Transform is a well-known method for finding lines. A detailed description of the Hough Transform can be found in "Digital Picture Processing", by Azriel Rosenfeld and Avinash C. Kak, (Academic Press, Inc. 1982) Vol. 2, pp. 121 -126. The Hough Transform converts an edge map image into a 2-D histogram with one dimension being the line orientation and the other being the line intercept. Hough Transform entry HT (x,y) represents the length of a line that has an orientation of x and an intercept of y. The edge orientation histogram H(x) can be obtained by manipulating the HT (x,y) histogram as follows:

H(x)=(ΣHT(x,y)²) 1 2

where the summation is over all y values.

The edge orientation (EO) algorithm is performed on the edge orientation histogram H(x) as follows:

EO = -ΣH(x) log H(x)

The software is sometimes manually guided, and parameters adjusted so that clusters of fibers and optionally curved fibers are properly measured. We refer to this procedure as an "automated" resolving process and thus this feature of the

system is referred to as an automated cluster resolving capability. See, also, United States Patent No. 6,985,628 to Fan, the disclosure of which is incorporated herein by reference in its entirety.

In general, the analysis is performed by setting a fiber diameter range and using Hough Transform analyses and rejecting results which are inconsistent with the physical image, discussed further below.

The system calibration is set using an NlST 25mm stage micrometer so that live tiling tolerance is as tight as possible. This is done by positioning the micrometer on the stage so that it lies where frames meet in the X and Y directions. Using the User Defined tiling method and gradient blend stitching option, set X, Y, and guard frame values so that tiling with the algorithm results in a calibration of +_20μm or less. (This is confirmed by imaging the stage micrometer.) Name the calibration file and set it as the System Calibration File.

The system operates by placing the Petri dish with dispersed fibers on the motorized stage and imaging an area approximately 65mm X 50mm. The specimen is then processed as follows:

(1) Switch on power to the automated system BEFORE launching ImagePro;

(2) Select Acquire and StagePro;

(3) Initialize stage by selecting the second radio button down (Use physical limits of stage...) and press continue on the following dialogue box. Wait for stage to stop;

(4) Click on the Lens/Mag page, select your calibration file from the drop down window. Click on Calibrate XY and then set from file. Choose the file name of the system calibration and click OK;

(5) Go to the Stage tab and set the imaging origin to the lower right corner - use preview to see where the camera is imaging. Click on set origin to current position;

(6) Set the Scan Area to 5x5 and save settings (including guard frame) with an appropriate name;

(7) Place a clean, empty sample dish on the stage and choose correct background and flat field radio buttons on the Acquire page;

(8) Acquire a background image (a single image). Select the image as the background image. Do not close the background image;

(9) switch to the sample dish and aquire the 5 X 5 sequence. Select

Processing from the menu bar and tile images from the drop down;

(10) Choose the sequence to input and select set from frames. Apply the user defined stitching method (previously established). Save mosaic;

(1 1) Launch Fasep with only one image open (Fasep performs functions on open images, only one image should be open at a time);

(12) Set measurement parameters. Length minimum = 0.5mm, fiber width is sample dependent;

(13) Set segmentation threshold so that the long fibers are adequately filled in without gaps (-205);

( 14) Perform Blob analysis;

(15) Set 3 bins (dust, single fiber, cluster) for classification using fiber width. Classify sample;

(16) Check classifications and modify if needed;

(17) Separate objects into the classes;

(18) After cluster images have been generated, choose one at a time for separation analysis;

(19) Initially begin with Min. Maxima Area set at 4 and Avg. Fiber Diameter set at 8 (for ~30 μm diameter fibers). Separate fibers using Hough Transform;

(20) Use Maxima click as cited in the Fasep directions when separation is unsuccessful. Increase the Avg. Fiber Diameter as appropriate to measure long or curved fibers;

(21) When all clusters are analyzed, open data collector and export data to clipboard.

This method enables appropriate pixel size calibration for short fibers and image processing so that approximately a 65 X 50mm area can be analyzed as shown in Figure 3. The process enables short and long fibers to be measured with accuracy. The fibers in the prescribed field are automatically imaged and measured so that sampling is unbiased.

In comparison, when applying more manual methods of measurement, the operator chooses the fibers for measurement and therefore sampling is more subjective. Also, long fibers may need to be centered within a field of view or imaging frame to be fully measured and therefore long fiber measurements are susceptible to operator bias. Repeatability of the measurements is limited to the agility and repeatability of the operator as well as pixel size. The automated imaging process ensures that fiber length is calculated using the same algorithms consistently. Accuracy (number of units per pixel) and imaging time were considered to arrive at a suitable magnification. The magnification instituted enables accuracy of + 2% for lmm fibers to + 0.2% for 10mm fibers. This allows imaging of a sample dish in less than 70 seconds. The number of fibers per sample dish will vary according to the extent of fiber length retention within the sample.

An easily performed and effective sample preparation method along with automated data collection enables collection of thousands of fiber lengths within minutes.

Another challenge is measurement of crossing or touching fibers. As the fibers in the sample become longer, the frequency of crossing and touching of long fibers increases. Most image analysis software tends to offer an editing capability that separates intersecting fibers by cutting them. The cut fiber segments are either measured, or excluded from the data. The end result is mismeasured or discarded fiber length data. These processes can have a significant impact on long fiber data. The method of the invention identifies crossing and touching fibers and measures each fiber separately and in its entirety.

In this method, fibers are identified as single fibers or clusters (touching fibers) as shown in Figure 4. The fibers are color-coded according to category (single fiber, or clustered fibers). The single fibers are measured and the touching fibers are individually grouped so that each cluster can be addressed separately.

Each cluster can be separated into individual fibers on a cases-by-case basis with the FASEP software. The software proposes an initial separation scheme to the operator. This may be accepted, modified if fiber separation can be improved, or rejected and tried again. Figure S shows a cluster of fibers. The software colors the boundary of each fiber so the operator can confirm that fiber length is fully measured. Once the measurements are accepted, the data is added to the previously collected fiber length measurements. There are a few cases, however, when fiber separation is not successful. A few examples of successful and unsuccessful separations are illustrated in Figure 6. An experienced operator is able to separate most clusters that initially show an unsuccessful separation scheme. The fibers illustrated are relatively straight, but experience has shown that a number of long fibers are gently curved. Preferably, Software is selected that accounts for fiber curvature.

As previously mentioned, repeatability of sample imaging and the measuring algorithms used is another benefit of the automated system. To demonstrate, one sample was imaged at the magnification established for LFRT analysis and fiber lengths were measured three consecutive times. The data is reported in Table 1.

T/US2007/078090

15

Table I: The identical sample was imaged automatically three times. Automation offers a tighter tolerance of imaging and measuring algorithms over manual methods which improves repeatability. Bin 1 equals fibers with a length less than lmm and greater than or equal to 0.5mm. The other bins correlate to fiber length similarly.

Bin Run 1 Run 2 Run 3 SW Dev

1 0 15 5 15 5 156 0 10

1 5 384 384 389 027

20 536 536 537 0 02

2 5 648 64 8 64 8 001

3 0 71 6 71 6 71 7 001

3 5 78 9 78 9 78 9 001

4 0 854 85 4 85 4 0 01

4 5 88 6 88 6 88 6 0 00

5 0 92 1 92 1 92 1 0 00

5 5 94 7 94 7 94 7 0 00

60 962 962 96 2 0 00

65 97 7 97 7 97 7 0 00

70 97 7 97 7 97 7 000

7 5 97 7 97 7 97 7 0 00

80 97 7 97 7 977 000

85 97 7 97 7 97 7 0 00

9 0 97 7 977 97 7 000

9 5 100 100 100 000

Count 251 250 251 AVG 0 02

Std

CV. 77 9% 77 7% 77 9% Dev 0 07

VoI Mean 2.42 2.42 2.42

To further confirm imaging and measuring repeatability, the sample was rotated 30°, 45°, 90°, and 120° from its original orientation, and data was collected at each angle. Table II shows the resulting measurements.

Table II: A sample was rotated to four positions and imaged automatically each time. Variability may be due to the calibration values of X and Y and/or operator intervention with cluster separation. Bin 1 equals fibers with a length less than 1 mm and greater than or equal to 0.5mm. The other bins correlate to fiber length similarly

Cumulative Volume Weighted Fiber Length

Bin Start 30 deg 90 deg 120 deg 180 deg

1.0 14 8 14.8 14 8 14.8 14.8

1.5 39.4 39.4 39.2 39.4 38.9

2 0 53 0 53 1 52.7 52.8 52.3

2.5 64.5 64.4 64.3 64.4 63 8

3 0 71 1 71 3 71.1 70.9 71.4

3.5 76 4 77.0 764 76.2 77.0

4 0 82.0 82 7 82.0 81 9 83.1

4.5 87 3 87.5 87.3 86.1 87.9

5.0 92.7 93 4 92 7 92 7 92 7

5 5 94.0 94.8 940 94.0 94.0

6.0 94.7 95.5 94 7 94 7 94.7

6 5 96 3 96 3 96.3 96.3 96.3

7.0 96.3 96.3 96.3 963 96.3

7 5 96 3 96 3 96.3 96.3 96.3

8 0 96.3 96.3 96.3 96 3 96 3

8 5 96.3 96.3 963 96 3 96.3

9 0 96.3 96 3 96 3 96.3 96.3

9.5 97.4 97.4 974 97 4 97 4

10 0 97 4 97.4 97.4 97.4 97.4

10.5 100 100 100 100 100

Max 10.4 104 104 10 4 10.4

Count 488 488 488 488 488

CV. 75 0% 74.5% 75 6% 75 6% 75.0%

VoI Mean 2 54 2 52 2.54 2 55 2 53 007/078090

17

Another important test parameter is the minimal number of fibers that should be measured to sufficiently represent a sample. Ten thousand fibers with lengths > 0.5mm were measured from a polypropylene sample believed to have moderate fiber length retention. The minimum fiber length incorporated in the data was arrived at based on terms presented in International Standard ISO 22314 (International Standard ISO 22314 "Plastics- Glass fibre reinforced products- Determination of fibre length "). and the work of Thomason and Vlug (Thomason, J. L₁ Vlug, M. A. et al, "Influence of fiber length and concentration on the properties of glass fiber-reinforced polypropylene: Part 1 - Tensile andflexural modulus ", Composites; 27 A; pp 477-484 (1996), "Part 4 - Impact properties ", Composites; 28A; p.p277-288 (1997)). . For this determination, the volume weighted mean was used. The preliminary results in Table III show that there is less change in the volume (length) weighted mean with a sample size over 3000 fibers. It is intended that the study be repeated several times to support the data.

Table HI: A comparison of sample size to volume weighted mean, shows that fluctuation in the mean decreases with a ssaamri ple size over 3000 fibers.

5

1000 2000 3000 4000 5000 6000 7000 8000 9000 10000

Max ΪTΪ ΪTS rTO VK2 ΪΪ3 ΪTo ϊϊl ΪTi ϊTo 115

Count 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000

CV. 101.2% 101.9% 101.0% 101.0% 97.8% 97.5% 99.7% 99.4% 100.0% 100.5%

VoMWIean 3.00 2.84 2.57 2.49 2.43 2.47 2.53 2.50 2.51 2.50

Informative data presentation and interpretation are fundamental to understanding fiber distribution within a molded part. The number distribution shows the range of fiber lengths in the sample. It is reported using a histogram illustrating frequency vs. fiber length, and cumulative percent. However, long fibers are usually fewer in number compared to short fibers but represent a more significant volume of the sample. Long fibers also contribute more substantially to some desired physical properties. As such, the number mean should never be reported alone, and should not be given too much weight when reviewing fiber length data, especially for LFRT. To arrive at more meaningful and representative expression of fiber length in a sample, the volume (length) weighted mean is calculated and reported. For example: given a sample with two 10mm fibers and 100 lmm fibers, number mean is about 1.1mm but the volume weighted mean is about 2.5mm. The long fibers are virtually ignored in the number statistics but are represented in the volume (length) weighted statistics.

The equation for calculating volume weighted data is shown below.

Although more representative of the sample than the numerical mean, the volume weighted mean is not sufficient to report alone. For example, in Figures 7 and 8 the volume (length) weighted mean is similar for both samples however the fiber length distribution is very different. Therefore distribution histograms and curves should be reported. Cumulative percents over 3 mm, 5mm and 8mm may be taken into consideration for further understanding and comparing data. The volume weighted coefficient of variance can also be a useful tool when drawing comparisons regarding variability between data sets. — Z_n, the mean fibre length, expressed in micrometres, calculated from the following equation: i = »

where

L_f ts the length of the ith fibre, n (s the number of fibres measured:

L₀, a weighted mean length, expressed in micrometres, calculated from the following equation:

L- - ^ where n_t is the number of fibres of length L-,

The method of the invention offers the ability to recover glass fibers from molded parts while retaining fiber pliability, therefore facilitating sample preparation and confidence in length retention during sample preparation. The automated data collection and measurement process reduced operator subjectivity and variability in length measurements encountered in a more manual procedure. The method enables length measurements of crossing and touching fibers in their entirety whereas most software edit options do not retain length of all crossing and touching fibers. The process is repeatable and accurate (+ 3% through + 0.3%), and hands-on time is less than other previously mentioned procedures. One preliminary conclusion is that fiber length analysis is most accurately and swiftly performed when the fibers are well- dispersed across the imaging dish and when touching fibers are minimized. Imaging several more sample dishes takes less hands-on time than editing and reviewing fiber clusters.

Increasing applications of LFRT materials warrant an effective method for quantifying fiber length distribution. This method will enable industries engaged in manufacturing LFRT parts to accrue data upon which a baseline for understanding long fiber length distribution can be formed, and improved upon with time and experience. Equipped with this ability, questions that have been posed since the genesis of LFRT materials regarding fiber length in molded parts can more adequately be addressed.

While the invention has been described in detail above, modifications within the spirit and scope of the present invention, set forth in the appended claims, will be readily apparent to one of skill in the art.

Claims

WHAT IS CLAIMED IS:

1. A method of measuring fiber length distribution in a long fiber reinforced thermoplastic composite comprising:

5 (a) recovering fiber from the composite by application of heat sufficient to liberate the fiber from matrix polymer; and

(b) dispersing the fiber on a substantially transparent imaging support;

] 0 (c) imaging the dispersed fiber in an automated imaging apparatus with an automated fiber cluster-resolving capability; and

(d) analyzing the image of the fiber dispersion including resolving fiber clusters into individual Fibers using an automated resolving 15 process.

2. The method according to Claim 1, including the further step of accepting or rejecting a proposed automated resolution of a fiber cluster.

20 3. The method according to Claim 1, wherein the imaging apparatus is adapted to produce an edge-map image and analyze the edge-map image utilizing Hough Transform analysis.

4. A method of measuring fiber length distribution in a long fiber reinforced 25 thermoplastic composite comprising:

(a) recovering fiber from the composite by application of heat sufficient to liberate the fiber from matrix polymer; and

(b) dispersing the fiber on a substantially transparent imaging support;

(c) imaging the dispersed fiber in an automated imaging apparatus having adjustable fiber diameter parameters as well as an automated cluster-resolving capability; and

(d) analyzing the image of the fiber dispersion including resolving fiber clusters into individual fibers using an automated resolving process, and adjusting fiber diameter parameters.

5. A method of measuring fiber length distribution in a long fiber reinforced thermoplastic composite comprising:

(a) recovering fiber from the composite by application of heat sufficient to liberate the fiber from matrix polymer while controlling the temperature so as not to embrittle the fibers;

(b) randomizing and dispersing the recovered fibers on a substantially transparent imaging support;

(c) placing the imaging support with dispersed fibers in an automated imaging apparatus;

(d) imaging the fiber dispersion;

(e) analyzing the image of the fiber dispersion with a length-calibrated image analyzer adapted to identify and size individual fibers of the dispersion in order to determine fiber lengths; and

(f) calculating a fiber length distribution based on an analysis of step (e).

6. The method according to Claim 5, wherein at least about 3,000 individual recovered fibers are analyzed for length.

7. The method according to Claim 5, including randomizing the fibers in a liquid medium to resolve clusters and remove other additives and residual polymer, and filtering the liquid medium to recover fibers.

8. The method according to Claim 7, wherein the liquid medium is an aqueous medium provided with an organic dispersant

9. The method according to Claim 8, wherein the organic dispersant is glycerin.

10. The method according to Claim 5, wherein the composite is a glass fiber/polyolefin composite and the image analyzer is calibrated to measure fiber length up to about 15mm.

11. The method according to Claim 10, wherein the image analyzer is calibrated to measure fiber length from about l/2mm to about 1 lmm.

12. The method according to Claim 5, including resolving fiber clusters into individual fibers.

13. The method according to Claim 12, including color coding fibers in order to confirm resolution of individual fibers in fiber clusters.

14. The method according to Claim 5, further comprising the steps of (i) cutting a test specimen from a longer sample and (ii) discarding fiber adjacent to cut surfaces prior to analyzing the fiber length distribution of the test specimen.

15. The method according to Claim 5, wherein the temperature is controlled to between about 400°C and 600°C when heating the composite.

16. The method according to Claim 15, wherein the temperature is controlled to be between about 425°C and 525°C when heating the composite.

17. The method according to Claim 5, wherein the composite is heated in a muffle furnace at a temperature of from about 425^°C to about 525°C for at least 6 hours.

18. A method of measuring fiber length distribution in a long fiber reinforced thermoplastic composite comprising:

(a) recovering fiber from the composite by applying heat sufficient to liberate the fiber from matrix polymer while controlling the temperature to be between about 400°C and about 600°C; and

(b) analyzing the recovered fiber for fiber length distribution using an automated imaging apparatus.