Experimental Studies of Fiber Surface Roughness Characterization by

Laser Backscattering

CHENG LUO and RANDALL R. BRESEE, Center for Materials Processing, 230 Jessie Harris Building, University of Tennessee,

Knoxville, Tennessee 37996-1900

Synopsis

The application of laser backscattering to the characterization of fiber surface roughness was studied experimentally using instruments based on optical Fourier transforms. Applications included comparing surface roughness of various fibers, measuring the effect of fiber delustering with TiOP and detecting surface crystallization. The optical transform method was found to be a simple means of providing nonintrusive, real-time measurements of fiber surface structure.

INTRODUCTION

The surface structure of fibers is important because of its effects on textile behavior during the processing of fibrous materials and performance of finished products. Prominent among surface characteristics is geometric roughness, be- cause it is closely related to frictional and optical properties. Frictional properties influence the basic mechanical behavior of textiles and fabric hand whereas optical properties affect the appearance of fibrous materials.

What is smooth and what is rough? Perhaps the most striking difference between smooth and rough surfaces is the fact that a smooth surface will reflect incident light specularly in a single direction whereas a rough surface will scatter incident light in numerous directions.'

In recent years interest in surface roughness research using Fourier transform methods has grown. In the late 1970s, B. J. Pernick' developed a method to measure the surface roughness of flat metallic objects with an optical Fourier spectrum analyzer. In 1987, Allardyce and George3 presented theoretical and experimental analysis of laser backscattering patterns from flat metallic sur- faces. They showed that the farfield backscattering pattern is the Fourier transform of the scattered object surface by comparing computer-calculated transforms with optical transforms measured with their Fourier spectrum an- alyzer. In the study reported here, we have applied the optical Fourier transform method to measure fiber surface roughness.

THEORY

Diffraction may be defined as any deviation of light from straight paths not explained by reflection or refra~t ion.~ As discussed previously, Fraunhofer

Journal of Polymer Science: Part B: Polymer Physics, Vol. 28, 1771-1780 (1990) Q 1990 John Wiley & Sons, Inc. CCC 0887-6266/90/01001771-010$04.00

1772 LUO AND BRESEE

diffraction is obtained when the distance between an aperture and observation plane is much greater than the aperture size. Then, the electromagnetic field a t an observation plane, E ( xo , yo), can be described by the Fraunhofer integral

where E ( x , y ) denotes the electromagnetic field at an aperture pl ne, k = 27r/ A is the wave number, X is the wave length, and Z is the distance between the fiber surface and the observation plane.

If spatial frequencies are defined as

u = x o / X Z and u =yo/XZ ( 2 )

the Fraunhofer diffraction expression is exactly the Fourier integral except for some extra multiplicative phase and amplitude factors outside the integral sign which can be grouped into a constant. The power of the illumination is not dependent on phase and therefore the intensity of a scattering pattern is a scaled version of the squared magnitude of the Fourier transform.

It has been noted that the Fraunhofer expression could be applied to back- scattering if the electromagnetic field at the aperture plane E ( x , y ) is replaced by a reflectance function, R ( x , y ) .

where A is a constant, and spatial frequencies have been inserted using eq. ( 2 ) . Thus, farfield scattering falls into the spatial frequency domain, and we can

obtain spatial frequency information from the Fourier spectrum. A perfectly smooth surface will show scattered intensity located only at zero frequency (for normal incidence of light source). On the other hand, a fiber surface which is not smooth will exhibit reduced zero frequency intensity while the scattered intensity at other frequencies will be increased according to the nature of the surface roughness. Therefore, the integral of all frequency components ratioed to zero frequency intensity will give a measure of fiber surface roughness without significant influence of diameter. A parameter called roughness coefficient, RC, was thus defined previously5 as

where lo is the intensity a t zero frequency and It represents the total intensities scattered at other frequencies. The It term is zero for a perfectly smooth surface and its RC value is equal to zero. The rougher a fiber surface becomes, the greater It becomes and RC will increase.

FIBER SURFACE ROUGHNESS 1773

EXPERIMENTAL METHODS

Two different experimental systems were employed in this work. A micro- computer-based instrument is shown schematically in Figure 1. Optical com- ponents were mounted on a graduated 2-meter optical rail. The light source was a 2 mW He : Ne laser with a wavelength of 0.6328 ym and beam diameter of about 1.5 mm. The sample holder permitted 360" rotation about the beam direction as well as translation in the vertical and transverse directions. Scat- tered radiation was projected onto a Mylar imaging screen located about 100 mm from the fiber sample. This screen was optically translucent and allowed the scattered image to be viewed from the rear of the screen by a silicon-vidicon camera. The signal output from the video camera was connected to an IBM personal computer equipped with a video digitizer interface card, 640 kB of user memory, two 360 kB floppy disk drives, an external 20 MB hard-disk drive, a pen plotter, and a dot matrix printer.

A single fiber was mounted on the rotatable sample holder and illuminated by the laser. The fiber area illuminated was determined by the laser beam and fiber diameters and ranged from 1.5 X 0.010 mm to 1.5 X 0.200 mm. After adjusting the camera for the optimum image, the scattering pattern was digitized and saved on the hard disk for data analysis. If desired, the scattering pattern could be plotted in either two or three dimensions. RC values were easily com- puted from the digitized scattering pattern.

A simpler and cheaper instrument also was employed. This instrument was somewhat similar to the computer-based system except the imaging screen, vidicon camera, computer, and plotter were replaced with either a 4 X 5 inch Polaroid film back and shutter or two inexpensive 5-mm diameter photodetectors and a voltmeter. The photodetectors were mounted on a two-dimensional po- sitioner. The upper phototransistor was positioned at zero frequency and par- tially covered with a mask so that a narrow slit of zero frequency light reached the detector but nonzero frequency components were stopped. The lower pho- totransistor was used to collect nonzero frequency components. Signals from each detector were amplified by uA741 operational amplifiers and light inten- sities were measured in volts. The Polaroid film holder was located in place of the photodetectors to photographically record scattering patterns using Type 55 P / N Polaroid film. The shutter positioned in front of the laser provided various exposure times during photographic exposure.

Sample Holder

n l a ser

1 Vidicon Camera

EL imaging Screen

IBM PC

Fig. 1. Microcomputer-based optical Fourier transform system.

1774 LUO AND BRESEE

RESULTS AND DISCUSSION

Measuring Fiber Surface Roughness

From the optical definition of smooth and rough, we know that light is re- flected specularly from a perfectly smooth surface, whereas rough surfaces scat- ter light away from the specular direction and reduce its intensity. This is shown in Figure 2, where the scattering patterns from fibers with four different levels of roughness were recorded by Polaroid film.

Figure 2 ( a ) shows the backscattering pattern of a single nylon fiber. Scat- tering from this fiber is predominantly specular, being intense, quite narrow, and aligned perpendicular to the fiber axis on the zero frequency line. This indicates the fiber surface is very smooth. Figure 2 ( b ) shows the backscattering pattern of a polypropylene fiber and slightly more nonzero frequency scatter can be seen than for nylon. Figure 2 (c ) shows the scattering pattern of a rayon fiber. Scattering can be seen to occur at a greater number of nonzero frequency components and indicates the fiber surface is rougher than nylon or polypro- pylene. Figure 2 ( d ) shows the scattering pattern of a wool fiber. The specular reflection (zero spatial frequency scatter) is greatly diminished by scattering at other frequency components. This indicates that the wool fiber surface is quite rough.

Fig. 2. Fiber backscattering patterns recorded on Polaroid films from fibers aligned vertically to the films. The ( 0 , O ) point is located approximately 8 mm off the left center of each photograph. ( a ) Nylon, ( b ) polypropylene, ( c ) rayon, and ( d ) wool fibers.

FIBER SURFACE ROUGHNESS

( a )

1775

Fig. 3. Three-dimensional plots of fiber backscattering patterns recorded with the vidicon- camera from fibers aligned parallel to the u axis. The ( 0 , O ) point is located at the center of the left edge of each plot. ( a ) Nylon, RC = 0.90, ( b ) polypropylene, RC = 1.84, (c ) rayon, RC = 4.20, and ( d ) wool, RC = 12.70.

Figure 3 shows single fiber backscattering patterns from the same samples recorded with the vidicon camera and plotted in three dimensions. Qualitative agreement between Figures 2 and 3 can be seen as the intensity a t u = 0 de- creased while high-frequency intensities increased with increasing fiber surface roughness. The RC values computed from these plots were 0.90,1.84,4.20, and 12.70 for nylon, polypropylene, rayon, and wool, respectively.

RC values were measured for five fibers using the simple measuring instru- ment with two photodetectors. Table I contains RC values that include the four fibers represented in Figures 2 and 3 as well as cotton. Comparing these results with those from the vidicon-camera-based system, the relative order of fiber roughness for the four fibers examined by both instruments is the same,

TABLE I Fiber Roughness Coefficient Values Measured With Two Photodetectors

Nylon Polypropylene Rayon Cotton Wool

RC 0.019 0.362 1.550 5.350 7.415

1776 LUO AND BRESEE

TABLE I1 Rayon Fiber Roughness Coefficient Values of 30 Specimens Measured With Two Photodetectors

Bright rayon Extra dull rayon

#2556 #2581 #4315 #4316

Average RC 2.522 2.877 3.899 3.599 SD 1.050 1.670 1.469 1.490 CV % 41.6 58.1 37.6 41.4

although absolute values of roughness coefficients are not identical. Differences in absolute values can be expected since different regions of scatter are detected and used for determining I. and It by each of the two instruments.

RC values were measured from four types of rayon using the simple instru- ment with two photodetectors. Results from measurements of 30 different spec- imens of each rayon type are summarized in Table 11. Mean RC values were found to be 2.522 and 2.877 for the bright rayons (#2556 and #2581) and 3.899 and 3.599 for the extra dull rayons (#4315 and #4316). Microscopic images showed that the variability among these fiber specimens was quite large and the large coefficients of variation for roughness coefficients reflects this vari- ability.

Student’s t tests were used to test for significant differences among the means. The results are presented in Table I11 for an a value of 0.05. A critical t value of 2.00 must be exceeded for any two means to be significantly different. Neither the two bright rayons nor the two extra dull rayons were found to be significantly different from each other. Although the t-test concluded that one of the bright rayons (#2556) was significantly different from the two extra dull rayons, one bright rayon (#2581) and one extra dull rayon (#4316) were concluded to be not different. The latter conclusion probably is affected by the large variation among fibers and might be changed if more fiber samples were analyzed.

Measuring the Effect of Delustering

Most delustering treatments involve deposition of particles in fibers to dif- fusely scatter light illuminating the fibers. A reduction in specular reflection

TABLE 111 “t” Test for Rayon Fiber RC Values

Bright rayon Extra dull rayon

#2556 #2581 #4315 #4316

#2556 - 0.130 4.106 3.182 #2581 0.130 - 2.474 1.737 #4315 4.106 2.474 - 0.772 #4316 3.182 1.737 0.772 -

(Y = 0.05 and critical t value = 2.00.

FIBER SURFACE ROUGHNESS 1777

and increase in diffise scatter lead to a reduction in fiber luster. The instruments used in this study might be used to measure delustering effects on single fibers.

A fiber sample set was prepared by mixing Ti02 particles in melted polyeth- ylene in concentrations of 0.0, 0.5, 1.0, 2.0, and 5.0%. Fibers with diameters of 40 pm were hand pulled from the melt. The relationship between RC values and Ti02 concentrations for these fibers is plotted in Figure 4, which shows that RC values increased with increasing addition of Ti02. The amount of TiOP existing on the surface of these fibers would be expected to increase as the amount of Ti02 added to the polymer melt increased. This conclusion was supported qualitatively by optical microscopy. Examples of scattering photo- graphs from this same fiber set are shown in Figure 5. The scattering patterns show that increasing concentrations of Ti02 result in increasing intensities of high-frequency scattering components. That is, fiber surface waviness shifts toward higher frequencies with increasing amounts of TiOz deposition.

Measurement of Crystallization

Polymers that have been crystallized under quiescent conditions may contain spherulitic crystalline structures. These roughly spherical structures can be detected by either polarized optical microscopy or small-angle light scattering in the forward direction. In this study, we investigated the use of optical back- scattering as another method of detecting crystallites. In contrast to polarized optical microscopy and forward light scattering, which provide information about crystallites in the bulk, the backscattering technique has the potential to provide information about surface crystallites. Measurement of surface crys- tallization would be particularly beneficial for understanding many processes, particularly those which presumably involve rapid surface cooling and conse- quent surface crystallization during processing such as melt blowing.

0 1 3 4 5

Ti02 Concentration Z Fig. 4. The effect of T,Oz concentrations on RC values.

1778 LUO AND BRESEE

(d) ( C )

Fig. 5. Backscattering patterns of 40 pm diameter polyethylene fibers with T,02 concentrations of ( a ) 0.5%, ( b ) 1.0%, ( c ) 2.0%, and ( d ) 5.0%. Fibers were aligned vertically.

The laser backscattering technique was applied to crystallization in this study by melting polyethylene, hand-drawing fibers from the melt, heating the fibers above their melting point on a glass plate, quickly mounting the glass

0 : I I I I

0 1 2 3 4

Time (second) Fig. 6. Roughness coefficient values of polyethylene during crystallization.

FIBER SURFACE ROUGHNESS 1779

(b) Fig. 7. Backscattering patterns of a polyethylene fiber ( a ) before and ( b ) after crystallization.

The fiber was aligned vertically.

plate on the instrument sample holder, and then allowing the polymer to cool under ambient conditions. Surface scattering was measured from the molten state through crystallization to the solidified state. Results are shown in Figures 6 and 7. RC data which are plotted in Figure 6 clearly show an abrupt increase, which occurred when the polymer crystallized. Relatively constant RC values were observed on both sides ofthe rapid change. Figure 7 shows typical scattering

1780 LUO AND BRESEE

patterns observed before and after crystallization. The increase in d i f i se scatter that accompanies crystallization is evident in this figure. Our measurements indicate that laser back scattering might be an effective way to detect surface crystallization.

SUMMARY AND CONCLUSIONS

Typical textile fibers have large surface-to-volume ratios, so their surface structures greatly influence many properties of fibrous materials. Prominent among surface structure is geometric roughness. The application of laser back- scattering to fiber surface roughness characterization was studied experimen- tally. In order to quantify roughness, a simple parameter called “roughness coefficient,” the ratio of scattered intensities to specular intensities, was used as a measure of surface roughness. Two instruments were employed to measure fiber surface roughness using laser backscattering. One involved detection of scattered light with a vidicon-camera and microcomputer, while the other in- strument employed only two simple photodetectors. Several applications of these instruments have been examined, including comparing surface roughness of various fibers, measuring the effect of fiber “delustering” with TiOz and detecting surface crystallization. In conclusion, the optical transform method employed here seems to have potential as a simple means of providing non- intrusive, real-time measurements of fiber surface roughness during processing or after use in service.

References

1. P. Beckmann and A. Spizzichino, The Scattering of Electromagnetic Waves from Rough

2. B. J. Pernick, Appl. Opt., 18, 796 (1979). 3. K. J. Allardyce and N. George, Appl. Opt., 26, 12 ( 1987). 4. J. W. Goodman, Introduction to Fourier Optics, McGraw-Hill, New York, 1968. 5. C. Luo and R. R. Bresee, J. Polym. Sci. Phys. Ed., 28, 1755 (1990).

Surfaces, Artech House, Nonvood, MA, 1987.

Received February 24,1989 Accepted November 1, 1989