**3. Results and discussions**

#### **3.1 Results of LFM analysis of fibers**

Images obtained in the AFM mode (topography) and in the LFM mode are presented in figures 4, 6, 7 & 8. The difference between the LFM backward and forward signals gives twice the friction force in volts. Friction coefficients were calculated by dividing the friction force (in nN) by the normal applied force (in nN), the latter being calculated and calibrated by using the contact force curve profile (see Fig. 5A), as described earlier. For each fiber, numerous tests were performed and the friction coefficient values were found to be reproducible when tests were carried out on different regions of a filament as well as on different filaments of the same glass fiber.

The friction coefficient values were found to be relatively small and varied from 0.01 to 2. These are relevant values, especially when we compare them to those obtained by other authors (Bhushan, 1985) who worked on polymeric films and other surfaces. As compared to film-film friction coefficient values for PET for example which are around 0.66-0.77, nanoscale friction coefficients of PET films measured by AFM/LFM are relatively small and are around 0.02-0.06.

#### **3.1.1 Desized glass fiber**

Glass fiber without size was equally analyzed by AFM/LFM. A desizing procedure was established, which consisted in heating the sized fiber at 600°C for 24 h so as to completely destroy the organic size. The topographical image (Fig. 4) shows that the desized fiber surface has a friction coefficient µ= 0.04 which stays constant all throughout the fiber

Static friction force values were obtained from the minimum values of the curve because, at the start, the microbalance platform was constrained to move upwards, and consequently static friction force values measured had negative values. So the signs of the friction force values were inverted. Moreover, before the vertical displacement of the platform, the balance was set to zero, while in practice, there was a load N at the extreme end of the

Capstan method generally applied for yarn-yarn friction measured by the F-meter was used

2 1 *<sup>T</sup> Ln T*

(3)

'T2' is the real static force measured. (-Fs+ N). The wrap angle '' is equal to the deflection angle '' of the vertical fiber for small values of deflection angle (see Fig. 3A). The deflection angle of the vertical fiber was calculated for each vertical displacement 'Z' of the horizontal

Images obtained in the AFM mode (topography) and in the LFM mode are presented in figures 4, 6, 7 & 8. The difference between the LFM backward and forward signals gives twice the friction force in volts. Friction coefficients were calculated by dividing the friction force (in nN) by the normal applied force (in nN), the latter being calculated and calibrated by using the contact force curve profile (see Fig. 5A), as described earlier. For each fiber, numerous tests were performed and the friction coefficient values were found to be reproducible when tests were carried out on different regions of a filament as well as on

The friction coefficient values were found to be relatively small and varied from 0.01 to 2. These are relevant values, especially when we compare them to those obtained by other authors (Bhushan, 1985) who worked on polymeric films and other surfaces. As compared to film-film friction coefficient values for PET for example which are around 0.66-0.77, nanoscale friction coefficients of PET films measured by AFM/LFM are relatively small and

Glass fiber without size was equally analyzed by AFM/LFM. A desizing procedure was established, which consisted in heating the sized fiber at 600°C for 24 h so as to completely destroy the organic size. The topographical image (Fig. 4) shows that the desized fiber surface has a friction coefficient µ= 0.04 which stays constant all throughout the fiber

vertical fiber. The real static (or dynamic) friction force measured was then (-Fs+ N).

to calculate the static friction coefficients during fiber-fiber motion (Gupta, 1993):

'T1 ' is the initial tension, that is the load applied to the vertical fiber + the fiber weight

**2.2.2 Static friction coefficients** 

fiber before calculating the friction coefficient

**3. Results and discussions** 

**3.1 Results of LFM analysis of fibers** 

different filaments of the same glass fiber.

are around 0.02-0.06.

**3.1.1 Desized glass fiber** 

surface. Fig. 5C shows that there is a great attraction of the tip by the sample before the former may be in contact with the fiber. This phenomenon may be explained by a higher surface energy of the clean desized glass fiber which is going to attract water molecules of the air very rapidly. Indeed, molecules of water form a film of water at the fiber surface, and this acts as a lubricant—hence, the very weak value of the friction coefficient.

Fig. 4. Desized glass fiber: (A) topographic image; (B) scope-mode forward and backward scanned AFM and LFM signals of section A–A'.

Fig. 5. (A) A normal contact force profile, (B) Contact force profile during scanning of fiber E and of (C) the desized glass fiber

Atomic Force Microscopy – For Investigating Surface Treatment of Textile Fibers 239

Furthermore, in the LFM mode, small fluctuations of the backward and forward signals were observed. Friction forces were determined by considering the average value of signals,

The glass fiber with the size A presents therefore, a physically and chemically heterogeneous surface, with bumps having a friction coefficient µ~0.04 while the overall surface has a friction coefficient µ~0.12. This would mean that during coating of glass fiber A by the starch size, the different constituents of the size are not distributed in a

When the topographic images were realized at a contact force greater than 70 nN, surface damage due to plowing of the size E by the AFM tip, was observed. A typical example of

Fig. 7. Topographic image of fiber E, E4 and scope-mode forward and backward scanned AFM and LFM signals of section A–A'for fiber E, and of section B–B'for fiber E4. (plowing

only.

homogeneous manner.

such topographic images is illustrated in Fig. 7A.

positions of fiber E are indicated by arrows)

**3.1.2.b Fiber E** 

#### **3.1.2 Sized glass fibers**

All sized glass fibers have very heterogeneous surfaces compared to the plain surface of the cleaned glass fiber.

#### **3.1.2.a Fiber A**

Typical images of fiber A, obtained in the AFM and LFM modes are illustrated in Fig. 6A and 6B, respectively. These have been obtained at a contact force of 38 nN. The topographic image in Fig. 6A, shows randomly distributed bumps of variable dimensions. The forward and backward scanned images in the LFM mode presented in Fig. 6 B, show that contrasts on the bumps are reversed when the scanning direction is changed. The scope mode AFM and LFM signals of the sections A-A' and B-B' corresponding to extremities of the topographic image of Fig. 6A are illustrated in Fig. 6C and 6D respectively. The scope mode LFM signals of the bumps reveal a friction force (Ff) of ~0.035 Volts (indicated by full lines), that is a friction coefficient µ ~0.04. The remaining surface (indicated by dashed lines) has a friction force two to three times greater, that is a friction coefficient µ~0.12.

Fig. 6. Fiber A: (A) topographic image; (B) forward and backward scanned LFM images; (C) scope-mode forward and backward scanned AFM and LFM signals of section A–A' and of section B–B' (D)

Furthermore, in the LFM mode, small fluctuations of the backward and forward signals were observed. Friction forces were determined by considering the average value of signals, only.

The glass fiber with the size A presents therefore, a physically and chemically heterogeneous surface, with bumps having a friction coefficient µ~0.04 while the overall surface has a friction coefficient µ~0.12. This would mean that during coating of glass fiber A by the starch size, the different constituents of the size are not distributed in a homogeneous manner.

#### **3.1.2.b Fiber E**

238 Atomic Force Microscopy – Imaging, Measuring and Manipulating Surfaces at the Atomic Scale

All sized glass fibers have very heterogeneous surfaces compared to the plain surface of the

Typical images of fiber A, obtained in the AFM and LFM modes are illustrated in Fig. 6A and 6B, respectively. These have been obtained at a contact force of 38 nN. The topographic image in Fig. 6A, shows randomly distributed bumps of variable dimensions. The forward and backward scanned images in the LFM mode presented in Fig. 6 B, show that contrasts on the bumps are reversed when the scanning direction is changed. The scope mode AFM and LFM signals of the sections A-A' and B-B' corresponding to extremities of the topographic image of Fig. 6A are illustrated in Fig. 6C and 6D respectively. The scope mode LFM signals of the bumps reveal a friction force (Ff) of ~0.035 Volts (indicated by full lines), that is a friction coefficient µ ~0.04. The remaining surface (indicated by dashed lines) has a

Fig. 6. Fiber A: (A) topographic image; (B) forward and backward scanned LFM images; (C) scope-mode forward and backward scanned AFM and LFM signals of section A–A' and of

friction force two to three times greater, that is a friction coefficient µ~0.12.

**3.1.2 Sized glass fibers** 

cleaned glass fiber.

**3.1.2.a Fiber A** 

section B–B' (D)

When the topographic images were realized at a contact force greater than 70 nN, surface damage due to plowing of the size E by the AFM tip, was observed. A typical example of such topographic images is illustrated in Fig. 7A.

Fig. 7. Topographic image of fiber E, E4 and scope-mode forward and backward scanned AFM and LFM signals of section A–A'for fiber E, and of section B–B'for fiber E4. (plowing positions of fiber E are indicated by arrows)

Atomic Force Microscopy – For Investigating Surface Treatment of Textile Fibers 241

increased crosslinking of the epoxy film former the lubricant is expurgated off onto the external surface in the form of blisters. This would have, consequently, the effect of

Results for measurements carried out under the following fixed conditions of constant speed (2 µm/sec), load (1 mg), relative humidity (45%), and temperature (25°C), are presented. Fig. 3B shows typical stick-slip curves for two different fibers A and E, obtained by plotting force as a function of the vertical displacement of the horizontal fiber. The stick-slip events for fiber E, are characterized by a greater average amplitude and a smaller frequency than those of fiber A. However, when each friction curve is considered in more detail, stick-slip irregularities along a filament as well as from one filament to another are observed. To obtain a 95% confidence level, friction measurements were carried out on four different

Fig. 9 shows the histograms of frequency distribution of static friction coefficients "µs" of fibers A and E respectively. For fiber E, the "µs" values follow a normal distribution with a mean value around 5 and a standard deviation of 1.7. However, for fiber A, a bimodal distribution of "µs" values is observed, with a mean value of the first distribution situated

Fig. 9. Histogram of static friction coefficients of fibers A and E, measured by electronic

The greater static friction coefficient values of fiber E (than the fiber A), during fiber-fiber friction can be explained by the greater frictional values detected by the LFM technique. However, as the LFM signals reveal no chemical heterogeneity of the fiber E surface (constant friction coefficient of 0.2), the great standard deviation of friction values during fiber-fiber friction can be attributed to the surface roughness of fiber E: this leads to different

**3.3 Comparison of fiber-fiber friction to friction analysis of surfaces by LFM** 

contact area each time a fiber is in contact with an another one.

**3.2 Fiber-fiber friction force measurements by the electronic microbalance** 

segments of 3 mm long filaments from five randomly chosen filaments.

around 2 and that of the second distribution around 6.

decreasing the coefficient of friction (µ = 0.03).

**Static friction coefficient (µs)** 

microbalance

Moreover, the contact force profile which gives the deflection signal as a function of the distance between the AFM tip and the fiber sample (as the piezodrive moves upwards and then downwards), shows that the AFM tip remained stuck to the size, and it was difficult to pull-off the tip from the fiber surface (Fig. 5B).

An example of topographic image of another fiber taken from the same roving of fiber E4, where plowing did not occur, is shown in Fig. 7B. The latter was made at a contact force of 62 nN and the contact force curve profile was normal. The difference between the forward and backward LFM signals revealed a constant friction coefficient of 0.2 throughout the fiber surface. This value is greater than that of the starch size on fiber A, for which the maximum friction coefficient is around 0.12.

The fact that some fibers are readily plowed by the AFM tip while others are not, may indicate a difference in the degree of cross-linking of the epoxy film-former of the size on fiber E, among fibers of a same roving.

When fiber E was annealed at 100°C for 60 h, the topographic image realized at 46 nN (Fig. 8A) shows no presence of plowing, and the contact force calibration curve profile was normal (see Fig. 5A). Nevertheless, it reveals an aggregation of matter with the formation of small blisters (indicated by arrows) of diameters of approximately 0.15 µm. LFM signals of a 1-µm² scanned area (Fig.8B) of a blister (full line) reveal a friction coefficient of µ= 0.03 which is smaller than that of the general surface of µ= 0.09.

Fig. 8. Annealed fiber E41: (A) topographic image; (B) scope-mode forward and backward scanned AFM and LFM signals.

According to the observations made on the annealed glass fiber E4, in both the AFM and LFM modes, it can be said that the increased crosslinking of the epoxy resin by annealing leads to a sized surface having a higher surface Young's modulus. The size is therefore less susceptible to be plowed by the AFM tip. One can also emit the hypothesis that during an increased crosslinking of the epoxy film former the lubricant is expurgated off onto the external surface in the form of blisters. This would have, consequently, the effect of decreasing the coefficient of friction (µ = 0.03).

#### **3.2 Fiber-fiber friction force measurements by the electronic microbalance**

Results for measurements carried out under the following fixed conditions of constant speed (2 µm/sec), load (1 mg), relative humidity (45%), and temperature (25°C), are presented. Fig. 3B shows typical stick-slip curves for two different fibers A and E, obtained by plotting force as a function of the vertical displacement of the horizontal fiber. The stick-slip events for fiber E, are characterized by a greater average amplitude and a smaller frequency than those of fiber A. However, when each friction curve is considered in more detail, stick-slip irregularities along a filament as well as from one filament to another are observed. To obtain a 95% confidence level, friction measurements were carried out on four different segments of 3 mm long filaments from five randomly chosen filaments.

#### **Static friction coefficient (µs)**

240 Atomic Force Microscopy – Imaging, Measuring and Manipulating Surfaces at the Atomic Scale

Moreover, the contact force profile which gives the deflection signal as a function of the distance between the AFM tip and the fiber sample (as the piezodrive moves upwards and then downwards), shows that the AFM tip remained stuck to the size, and it was difficult to

An example of topographic image of another fiber taken from the same roving of fiber E4, where plowing did not occur, is shown in Fig. 7B. The latter was made at a contact force of 62 nN and the contact force curve profile was normal. The difference between the forward and backward LFM signals revealed a constant friction coefficient of 0.2 throughout the fiber surface. This value is greater than that of the starch size on fiber A, for which the maximum

The fact that some fibers are readily plowed by the AFM tip while others are not, may indicate a difference in the degree of cross-linking of the epoxy film-former of the size on

When fiber E was annealed at 100°C for 60 h, the topographic image realized at 46 nN (Fig. 8A) shows no presence of plowing, and the contact force calibration curve profile was normal (see Fig. 5A). Nevertheless, it reveals an aggregation of matter with the formation of small blisters (indicated by arrows) of diameters of approximately 0.15 µm. LFM signals of a 1-µm² scanned area (Fig.8B) of a blister (full line) reveal a friction coefficient of µ= 0.03

Fig. 8. Annealed fiber E41: (A) topographic image; (B) scope-mode forward and backward

According to the observations made on the annealed glass fiber E4, in both the AFM and LFM modes, it can be said that the increased crosslinking of the epoxy resin by annealing leads to a sized surface having a higher surface Young's modulus. The size is therefore less susceptible to be plowed by the AFM tip. One can also emit the hypothesis that during an

pull-off the tip from the fiber surface (Fig. 5B).

friction coefficient is around 0.12.

scanned AFM and LFM signals.

fiber E, among fibers of a same roving.

which is smaller than that of the general surface of µ= 0.09.

Fig. 9 shows the histograms of frequency distribution of static friction coefficients "µs" of fibers A and E respectively. For fiber E, the "µs" values follow a normal distribution with a mean value around 5 and a standard deviation of 1.7. However, for fiber A, a bimodal distribution of "µs" values is observed, with a mean value of the first distribution situated around 2 and that of the second distribution around 6.

Fig. 9. Histogram of static friction coefficients of fibers A and E, measured by electronic microbalance

#### **3.3 Comparison of fiber-fiber friction to friction analysis of surfaces by LFM**

The greater static friction coefficient values of fiber E (than the fiber A), during fiber-fiber friction can be explained by the greater frictional values detected by the LFM technique. However, as the LFM signals reveal no chemical heterogeneity of the fiber E surface (constant friction coefficient of 0.2), the great standard deviation of friction values during fiber-fiber friction can be attributed to the surface roughness of fiber E: this leads to different contact area each time a fiber is in contact with an another one.

Atomic Force Microscopy – For Investigating Surface Treatment of Textile Fibers 243

Globally therefore, there can be four different types of contact during the fiber-fiber friction. We can compare these configurations to the results obtained by Yamaguchi (Yamaguchi, 1990) for polymer-polymer and polymer-steel friction force measurements. Yamaguchi observed that the friction coefficient of any polymer against PTFE (polytetrafluoroethylene) was nearly the same and of a small value (µ~0.1), while the friction of a polymer against steel varied a lot. In fact, it is the polymer presenting a weak friction coefficient that imposes the value of the relative friction coefficient (in the limit of the study considered). Therefore it can be concluded from our study that contacts a, b and c lead to a relative weaker friction coefficient than contacts of the type d. This would lead to a bimodal distribution of the relative friction coefficients during fiber-fiber friction and would explain our experimental

In light of the above results and discussions, both techniques of measuring friction forces, by Lateral force microscopy and an electronic microbalance, seem to be invaluable methods for characterizing frictional properties of sized glass fibers. The AFM/LFM successfully determines the topography and chemical nature of sized and desized glass fibers. Sized fibers have both physical and/or chemical heterogeneities while the desized bare glass fiber

AFM/LFM results also help to better understand friction force results obtained at a larger macro scale, particularly the widespread values of friction coefficients during fiber sliding. Nanosccale friction values by AFM/LFM are smaller than 'micro' friction values during fiber-fiber friction because the nature and the area of contact are different in both cases.

**Part II: Air-atmospheric plasma treatment of PET (Polyethylene Terephtalate)** 

Polyester fabrics made from PET poly(ethylene terephtalate) account for almost 50% of all fiber materials. PET fibers have high uniformity, mechanical strength or resistance against chemicals or abrasion. However, high hydrophobicity, the build-up of static charge, stain retention during laundering and being difficult to finish are undesirable properties of PET. Enhancement of the hydrophilicity of PET fibers is a key requirement for many applications, ranging from textile production to applications in the biomedical field. In the textile field, increased hydrophilic properties improves comfort in wear with a better moisture management due to increased wettabilility, wicking, adhesion to other materials (i.e. coating), and dyeing (Pastore M, 2001). Several strategies can be adopted to increase the surface energy and hence the hydrophilicity of PET fibers such as by chemical finishing or grafting, chemical surface treatment with NaOH (Collins, 1991, Haghighatkish, 1992), biochemical treatment with enzymes (Vertommen, 2005) or physical surface treatment using plasma. Treatment with NaOH is environmentally unfriendly and causes drastic weight and strength losses (Collins, 1991), while certain plasma treatments would be of interest from an

**woven fabrics studied by atomic force microscopy** 

results.

**4. Conclusions** 

**1. Introduction** 

environmental point of view.

is completely plain and smooth.

In the case of fiber A, in addition to surface roughness, chemical heterogeneity due to two regions with different friction coefficients (0.04 and 0.12) was observed by LFM. The great standard deviation of "µs" as well as its bimodal distribution, during fiber-fiber friction are thus due to surface roughness and chemical heterogeneity.

The low friction coefficient values obtained by LFM may be explained by the fact that sliding of AFM tip on a fiber sample surface is easier (µ < 1), than that of a fiber on another fiber surface (µs~1-10). Moreover in LFM, the friction force analyzed is a dynamic one, while in this case, it is the static friction coefficient which is being evaluated. The friction coefficient values disparities can be related to scale difference of measurements and to the nature and surface of contact which are different in both cases. In AFM, a Si3N4 tip having a radius of curvature of about 30 nm, is in contact with the fiber surface while during fiberfiber friction measurements sized fibers of diameter 11 µm are in contact.

## **3.4 Determining the theoretical contact area during the fiber-fiber friction**

The theoretical contact area during friction, can be calculated by using the DMT and JKR theories of contact mechanics. But according to Pashley, as the glass fiber is rigid and the fiber radius as well as its surface energy small, the DMT theory should be applied (Pashley, 1984). For orthogonally placed fibers, if contact occurs at one point, the contact radius derived by the DMT is 0.118 µm . However, contact between the two fibers occurs on a length of a circular arc of 2r (/360) (where the wrap angle =2.3° and the fiber radius: r=5.5 . 10-6 m (see Fig. 3). Thus the area of contact 'A' between the two fibers is approximately an ellipse, and it is evaluated to be A=0,12 µm² (details of this result is published in another paper ( N. Behary et al., 2000).

#### **3.5 Types of contact possible during the fiber-fiber friction of the fiber A**

The AFM/LFM measurements of the fiber A revealed two regions of distinct friction coefficient (µ~0.04 and 0.12 respectively), each of them having a surface area comprised between approximately 0.1 and 1µm², that is nearly the same contact area as that during fiber-fiber friction. Thus, the main types of contacts that can take place would have the configurations illustrated in Fig. 10.

Fig. 10. Modelling the different contacts possible during fiber-fiber friction of the sized fiber A (the friction coefficients are those evaluated by LFM)

Globally therefore, there can be four different types of contact during the fiber-fiber friction. We can compare these configurations to the results obtained by Yamaguchi (Yamaguchi, 1990) for polymer-polymer and polymer-steel friction force measurements. Yamaguchi observed that the friction coefficient of any polymer against PTFE (polytetrafluoroethylene) was nearly the same and of a small value (µ~0.1), while the friction of a polymer against steel varied a lot. In fact, it is the polymer presenting a weak friction coefficient that imposes the value of the relative friction coefficient (in the limit of the study considered). Therefore it can be concluded from our study that contacts a, b and c lead to a relative weaker friction coefficient than contacts of the type d. This would lead to a bimodal distribution of the relative friction coefficients during fiber-fiber friction and would explain our experimental results.
