Structure analysis of needle-punched nonwoven fabrics by X-ray computed tomography

Abstract

Needle-punching conditions determine the structure of nonwoven fabrics and the structure determines their tensile properties. However, the structural parameters of nonwoven fabrics and the relationship between these parameters and tensile properties have not been quantitatively analyzed. Therefore, we analyzed the structure of needle-punched nonwoven fabrics by X-ray computed tomography (XCT). The relationships between the needle-punching conditions, tensile properties, and structural parameters, such as fiber-volume fraction and three-dimensional fiber orientation, were investigated. The fiber-volume fraction in the middle layer of the fabric was clearly larger than that of the bulk above a compression ratio of 1.4. With increasing needle penetration depth, the fibers tended to become oriented both parallel and perpendicular to the normal direction of the fabric plane while avoiding the intermediate direction. A linear relationship was found between the obtained volume fraction of fibers oriented in the normal direction and the tensile strength of the fabric. These results demonstrate that XCT image analysis is effective to evaluate the structure of needle-punched nonwoven fabrics and to design the properties of nonwoven fabrics.

Keywords

needle-punched nonwoven fabric X-ray computed tomography fabric structure fiber-volume fraction fiber orientation tensile strength

In the needle-punching process, the barbs of a needle catch the fibers and force them along the normal direction (ND) of the fabric-plane.^1–4 Because the structure of needle-punched fabrics is developed by the fiber movement, the structure of the nonwoven fabric obtained is determined by the needle-punching conditions.² There are numerous reports on the relationship between the needle-punching conditions and properties of obtained fabrics, but few of these studies consider the structure of the obtained fabric.^5,6 Hearle and Sultan⁷ and Watanabe et al.⁸ evaluated the state of the needle-punched fibers by measuring the penetration force applied to the needle and counting colored (tracer) fibers in the needle-punched fiber bundle.^2–4 In the latter research, they categorized the structures of the fiber bundles into loop and pillar structures. If the needle-punched fibers did not reach to the bottom of the fabric, a loop structure formed. Such a structure tended to form when the penetration depth of the needle was small. In contrast, the pillar structure formed when the needle-punched fibers reached the bottom of the fabric, which tended to be observed at large needle penetration depth. The needle-punched fiber bundles were connected to each other by bridging fibers.^9,10 Moreover, these fiber bundles passed through the bottom of the fabric and were connected to each other along the machine direction (MD) of the fabric, making a stitch structure.^1,3 These structures are important because along with the bridging fibers they are thought to bear external force applied to the fabric. External force is also applied to the loop or pillar structures through the bridging fibers or stitch structure, and they bear the force of the friction between needle-punched fibers. Therefore, the tensile properties of fabrics are decided by their structure. However, Hearle and Purdy,² Hearle and Choudhari,³ Miao⁴ Hearle and Sultan,⁷ and Watanabe et al.⁸ did not analyze the overall structure of the fabric but focused only on the structure of the needle-punched fibers, so their findings cannot be applied to the quantitative design of tensile properties. For this purpose, it is desirable to conduct an overall structure analysis of needle-punched nonwoven fabrics.

Because needle-punched nonwoven fabrics have a three-dimensional (3D) complex structure, 3D structure analysis is suitable to analyze their overall structure. Recently, X-ray computed tomography (XCT) has started to be used for the 3D structure analysis of materials. XCT allows nondestructive observation and quantitative analysis of the internal microstructure of materials. XCT has been used to analyze the structure of nonwoven fabrics, including needle-punched nonwoven fabrics.^11–18 Gilmore et al.¹¹ first applied XCT to observe the 3D structure of nonwoven fabrics. Following their research, several research groups have analyzed the structure of nonwoven fabrics by XCT, and some have attempted to reveal the relationship between the structure and properties. For example, Manickam and McCutcheon¹² focused on the void structure of nonwoven fabrics, that is, they analyzed the pore diameter and porosity. Soltani et al.¹³ also analyzed the structure of nonwoven fabrics by XCT. They investigated the influence of fiber orientation on the permeability of nonwoven fabrics. Based on these structure analyses, Jeon et al.¹⁶ applied XCT structure analysis to the field of needle-punched nonwoven fabrics. They evaluated the in-plane fiber orientation, fiber-volume fraction, and contact level between fibers of needle-punched nonwoven fabrics, and analyzed the change in these structure parameters during tensile deformation. However, the 3D fiber orientation has not been analyzed. The effect of needle-punching conditions on the structure parameters has not been studied by XCT either. Moreover, the relationship between the structure and properties of obtained fabrics has not been determined quantitatively. Therefore, we use XCT to analyze the microstructure of needle-punched nonwoven fabrics in this study. In this analysis, the volume fraction and orientation of fibers are evaluated as structural parameters of the fabric. The relationship between these structural parameters and the needle-punching conditions, including needling density and needle penetration depth, is investigated. Moreover, the relationship between the structural parameters and tensile strength, one of the most important properties of fabrics, is also evaluated.

Methods

Sample preparation

We used crimped short fibers of polyethylene terephthalate with a circular cross-section. The fiber diameter and length are 40 ± 3.3 µm and 51 mm, respectively. Tufts of the fibers were opened by an OP-300 opener, and subsequently carded by an SRC-400 carder (Takeuchi Seisakusho & Co, Ltd, Osaka, Japan) to produce parallel-laid webs with a basis weight of 200 g/m².

We used an NL-500 needle-punching machine (Takeuchi Seisakusho & Co, Ltd) and FPD-1-40 needles (Organ Needle Co., Ltd, Nagano, Japan), which had a regular barb spacing¹⁹ and barb depth of 110 µm, as shown in Figure 1. We planted 960 needles in a needle board with a width of 480 mm and length of 240 mm (37 or 38 needles to a row and 25 needles per column). The web was fed at 7, 4, or 2 mm/punch, corresponding to a needling density of 20, 40, or 80 punches/cm², respectively. The penetration depth of the needle was set at 6.4, 12.7, or 19 mm, corresponding to the first, fourth, and seventh barbs reaching the top surface of the bedplate. Nine samples of needle-punched nonwoven fabrics produced under different conditions and a web as a reference were obtained and analyzed.

Figure 1.

Schematic image of a needle. The cross-sectional shape is triangular. Distances between the needle point and the barbs are described.

The basis weight (bw), thickness (t_bulk), and tensile properties of the obtained samples were measured. The basis weight was measured by weighing five samples with dimensions of 100 mm × 100 mm. The thickness was measured according to ISO 9073-2. From the obtained basis weight and thickness, we calculated the fiber-volume fraction ( $Φ_{bulk}$ ) by the following equation and by assuming a fiber density (ρ) of 1350 kg/m³

Φ_{bulk} = \frac{bw}{ρ t_{bulk}}

(1)

The breaking load per unit width (N/m) and elongation were measured in the MD and cross-machine direction (CD) according to ISO 9073-3. The tensile strength (N/tex) was calculated from the obtained breaking load per unit width divided by the basis weight of each sample.^1,3

XCT observation

Three samples for each set of needle-punching conditions were investigated by XCT. Each sample with dimensions of 20 mm × 20 mm was placed on a sample stage, which consisted of a pair of acrylic plates and fixing tools. Because sample thickness was easily affected by handling, a pressure of 120 Pa was applied to the sample by the acrylic plates for 30 s to normalize sample thickness, and then the plate was fixed at that height.

We placed the sample stage on a Skyscan 1272 XCT device (Bruker, MA, USA). X-rays generated at 50 kV and 200 µA were used. A series of 961 transmission images with 2452 × 1640 pixels (5 µm/voxel) was obtained by rotating the sample around the ND axis by 0.2°/image. From the series of transmission images, a sequence of two-dimensional (2D) tomographic images (in gray-scale with 256 gradations) was reconstructed.

In addition to the above-described thickness measurement according to ISO 9073-2, we also measured the thickness of samples in XCT images, which was defined as the distance between the pair of acrylic plates in tomographic images. Compression ratio λ was calculated by dividing the web thickness by that of each sample.

Results and discussion

Properties of needle-punched samples

The properties of needle-punched and web samples are listed in Table 1. The basis weight and thickness decreased with increasing needle penetration depth and needling density. The decrease of basis weight is caused by the expansion of the web induced by the spreading of fibers by the needle.¹ Conversely, the decrease of thickness is mainly a result of compression of the web structure along the ND. As the fibers on the top surface of web are punched into the interior by the needle, the web is compressed by the frictional force between the needle-punched fibers.¹ However, when the penetration depth of needle was 6.4 mm, the sample thickness did not change much from that of the web, and it was hardly affected by needling density. Because of the slight compression and decrease in basis weight, the fiber-volume fraction decreased from that of the web at a penetration depth of 6.4 mm. The number of needle-punched fibers decreases with the number of barbs that penetrate into the web,^2,18 so the frictional force acting between the needle-punched fibers should also decrease. In addition, because the most of the needle-punched fibers did not reach the bottom of the web when the penetration depth was 6.4 mm, it was difficult to form the stitch structure.^1,3,18 Therefore, the thickness did not change much when the penetration depth was 6.4 mm.

Table 1.

Properties of needle-punched samples

Penetration depth of needle^a (mm)	Needling density (punches/cm²)	Basis weight (bw) (g/m²)		Thickness (t_bulk) (mm)		Fiber-volume fraction (Φ_bulk) (%)	Tensile strength × 10^–5 (N/tex)				Elongation (%)
		Basis weight (bw) (g/m²)		Thickness (t_bulk) (mm)			MD		CD		MD		CD
		Ave.	S.D.	Ave.	S.D.		Ave.	S.D.	Ave.	S.D.	Ave.	S.D.	Ave.	S.D.
Web		197	12	8.14	0.67	1.79	4.3	0.4	1.4	0.2	38	6	68	9
6.4	20	189	13	8.36	0.46	1.67	14.1	2.3	3.0	0.6	43	4	90	24
	40	178	9	7.68	0.41	1.71	17.5	2.1	3.2	0.6	49	5	66	15
	80	171	7	7.47	0.18	1.69	21.4	1.6	4.7	0.9	45	4	87	13
12.7	20	172	2	6.89	0.17	1.85	29.5	2.8	5.2	0.5	53	5	117	13
	40	158	4	5.97	0.12	1.95	41.8	3.6	10.1	2.1	51	7	119	36
	80	160	4	6.13	0.12	1.93	53.4	6.9	13.0	1.8	63	10	172	13
19	20	158	5	6.00	0.14	1.90	33.5	1.7	7.2	0.3	57	4	158	13
	40	144	6	5.47	0.24	1.94	50.5	2.4	14.7	1.7	75	3	188	27
	80	145	5	4.90	0.18	2.18	86.6	18.4	32.8	4.8	99	13	176	27

Penetration depth of needle at 6.4, 12.7, or 19 mm that corresponds to first, fourth, and seventh barbs that reach the bottom surface of the web, respectively.

CD: cross-machine direction; MD: machine direction.

The tensile strength along the MD of the fabric was larger than those in the CD for all conditions, while the elongation at maximum tensile strength showed the opposite tendency. The tensile strength in both the MD and CD increased with needle penetration depth and needling density, while the elongation increased only with needle penetration depth. The directional dependency of tensile properties is related to the fiber orientation in the original web; that is, the fibers are preferentially oriented in the MD rather than the CD because the web consisted of fibers laid parallel. Meanwhile, because the elongation and thickness tendencies correlated well, it is thought that the strength was decided by not only the state of the needle-punched fibers but also the number of needle-punched fiber bundles, which is proportional to the needling density.

XCT image analysis

Representative 3D and 2D tomographic images of the samples are shown in Figures 2 and 3. Both images were binarized at a certain threshold level of brightness to eliminate most noise.¹⁸ The tomographic image was obtained parallel to the fabric surface at the half thickness of the fabric, as illustrated by the dashed line in Figure 2. Three sets of 3D images were obtained for each set of needle-punching conditions, and thus three tomographic images were used to analyze each sample.

Figure 2.

Three-dimensional image of a needle-punched nonwoven sample obtained with a penetration depth of 12.7 mm and needling density of 40 punches/cm². CD: cross-machine direction; MD: machine direction.

Figure 3.

Binarized tomographic image obtained at the half thickness of a needle-punched sample with a penetration depth of 12.7 mm and needling density of 40 punches/cm².

Objects of the fiber cross-section were extracted from the tomographic image by assuming an ellipsoidal cross-section. We focused on objects inside a region of interest (ROI) with a 10-mm diameter, as described in Figure 3. The major (a) and minor (b) diameters of the objects were obtained by image analysis. The b distribution of the obtained objects is shown in Figure 4. Most objects are thought to be fiber cross-sections, but some are noise. Moreover, some fiber cross-sections were detected incorrectly because they could not be fitted well by the ellipsoid. Two types of incorrectly detected objects exist, that is, a combined object of two or more neighboring fibers, and a long-curved object, which is the cross-section of the fiber that is aligned almost parallel to the scanned surface. Therefore, we considered objects with a b of 30–50 µm as the correctly detected single-fiber cross-section because the fiber diameter (40 µm) should be identical to b of the object if it is a fiber. Objects with a b of less than 30 µm can be considered to be noise.¹⁶ Conversely, most objects with a b that exceeds 50 µm are thought to be fiber cross-sections that were detected incorrectly. The number fraction of the objects of 30 µm ≤ b ≤ 50 µm to the objects of 30 µm ≤ b and the averaged b of the selected objects are listed in Table 2. Numerically, at least 83% of objects were considered to be correctly detected single-fiber cross-sections. The averaged b for all samples was almost the same at 43.4–44.4 µm, which is ∼ 10% larger than the actual fiber diameter. A similar averaged diameter ensures that most analyzed objects were fitted properly by the ellipsoid. However, a somewhat larger averaged b and incorrectly detected objects are remaining problems. They affect the fiber-volume fraction and orientation distribution, so the effects are discussed in the following sections.

Figure 4.

Distribution of the minor diameter of objects obtained from the region of interest in Figure 3.

Table 2.

Structural parameters of needle-punched samples obtained by X-ray computed tomography image analysis

Penetration depth of needle^a (mm)	Needling density (punches/cm²)	Compression ratio λ	Number fraction of correctly detected objects (%)	Averaged minor diameter of selected objects (µm)	Fiber-volume fraction (Φ_XCT) (%)		Second-order orientation factor
					Fiber-volume fraction (Φ_XCT) (%)		f_MD	f_CD	f_ND
					Ave.	C.V.	f_MD	f_CD	f_ND
Web		1.00	88	44.3	1.69	8%	0.14	0.11	–0.25
Compressed web		2.24	83	43.8	2.77	–	0.15	0.19	–0.34
6.4	20	0.98	88	44.3	1.46	22%	0.08	0.08	–0.15
	40	1.33	86	44.4	1.73	6%	0.09	0.07	–0.16
	80	1.37	87	44.2	1.75	9%	0.07	0.05	–0.12
12.7	20	1.33	90	44.2	1.84	10%	0.16	0.03	–0.18
	40	1.55	86	44.0	2.24	4%	0.16	0.01	–0.17
	80	1.60	86	43.8	2.14	21%	0.12	0.02	–0.14
19	20	1.49	88	43.8	2.18	10%	0.13	0.05	–0.18
	40	1.63	85	43.9	2.58	13%	0.12	0.04	–0.17
	80	1.93	85	43.4	2.96	6%	0.11	0.04	–0.15

Penetration depth of needle at 6.4, 12.7, or 19 mm that corresponds to the first, fourth, and seventh barbs that reach the bottom surface of the web, respectively.

Fiber-volume fraction

The fiber-volume fraction of the samples was calculated with the objects assigned as scanned fiber cross-sections using the following equation

Φ_{XCT} (%) = \frac{V_{f}}{V} = \frac{\sum_{i} a_{i} b_{i}}{D_{ROI}^{2}}

(2)

where Φ_XCT is the fiber-volume fraction, V is the volume of the scanned ROI, and D_ROI is the diameter of the ROI, which was 10 mm, as shown in Figure 2. V_f is the total volume of fibers included in the scanned ROI, and was calculated using a and b of each object (i).

The obtained volume fractions are listed in Table 2. The coefficient of variation for all volume fractions was less than 22%, confirming the reproducibility of the results. Figure 5 shows the fiber-volume fraction obtained by XCT ( $Φ_{XCT}$ ) plotted against the compression ratio of thickness for each sample. The volume fraction obtained by bulk measurements ( $Φ_{bulk}$ ) is also plotted. There was no clear difference between $Φ_{XCT}$ and $Φ_{bulk}$ below the compression ratio of 1.4, while the $Φ_{XCT}$ clearly increased more than $Φ_{bulk}$ over the compression ratio. While $Φ_{bulk}$ indicates the averaged value of the overall fabric, $Φ_{XCT}$ reveals the particular value in the middle layer of the fabric. Therefore, the above results show that the fiber fraction in the middle layer of the fabric is clearly higher than that of the surface layers above a compression ratio of 1.4. A clear density gradient along the ND develops above a certain compression ratio.

Figure 5.

Relationship between the compression ratio and fiber-volume fraction obtained by bulk measurements and tomographic imaging.

To ensure the above presumption, the distribution of $Φ_{XCT}$ along the ND is shown in Figure 6. Here, $Φ_{XCT}$ was measured by tomographic images of every 500 µm interval. The $Φ_{XCT}$ values in the middle layers of the fabric were larger than those of the surface layers above the compression ratio of 1.4, whereas they were almost uniform below 1.4. The averaged $Φ_{XCT}$ and $Φ_{bulk}$ are also shown on the left-hand side of Figure 6. They showed almost the same trend in the compression ratio. However, the averaged Φ_XCT was slightly smaller than Φ_bulk. This result should be caused by a removal of objects with a b that exceeded 50 µm, despite most of them being thought to be a fiber cross-section. If we consider a ∼10% larger value of averaged b than the actual fiber diameter, 20–30% of averaged Φ_XCT values were thought to be reduced by the removal. The assumed reduction in volume fraction corresponds to approximately twice the number fraction of incorrectly detected objects, as shown in Table 2. To obtain a more accurate Φ_XCT, further studies on the image-processing conditions, the threshold level of the binarization, and the treatment of removed objects, for example, are expected.

Figure 6.

Distribution of the fiber-volume fraction along the fabric normal direction. P and D denote the needle penetration depth (mm) and needling density (punches/cm²), respectively. “Ave. Φ_XCT” data represent the averaged Φ_XCT throughout all measured layers. Volume fraction $Φ_{bulk}$ was obtained from the basis weight and thickness, as listed in Table 1.

Fiber orientation distribution

The fiber orientation was characterized by the distribution along the inclination angle (θ) from the ND, as illustrated in Figure 7. The cosine of θ was obtained by image analysis using equation (3) ^16,17,20

cos θ = \frac{b}{a}

(3)

Figure 7.

Definition of fiber orientation angles ϕ and θ, where a and b are the major and minor diameters of scanned fiber cross-sections, respectively. CD: cross-machine direction; MD: machine direction; ND: normal direction.

The obtained fiber orientation distribution in the web is shown in Figure 8. The vertical axis indicates the probability obtained from equation (4)

Probability = \frac{V_{f (cos θ)}}{V_{f}}

(4)

where V_f_(cos _θ ₎ indicates the total volume of fibers oriented along each block of cosθ. The probability here is not the probability of the number of fiber cross-sections but the probability of fiber length. The probability of cosθ being less than 0.1 should be decreased by the removal of objects of b that exceed 50 µm. As mentioned in the previous section, many long-curved objects were included in the removed objects. Most were regarded as the cross-section of a fiber that is aligned almost parallel with the scanned surface, but they were removed because their inclination angle could not be determined by the ellipsoid fitting.

Figure 8.

Fiber orientation distribution (θ) of the web and its dependence on the compression ratio.

As shown in Figure 8, most fibers in the web were oriented perpendicular to the ND. This result is reasonable because the sample is a parallel-laid web. However, some fiber cross-sections were oriented along the ND. The ND-oriented cross-sections are thought to be observed by fiber crimping. Because the fiber orientation was analyzed by the cross-sectional shape in this study, a short-range fiber axis orientation was obtained. In general, the long-range orientation is higher than the short-range orientation because of fluctuation in the direction of the fiber axis, such as from fiber crimping. That is to say, the long-range fiber axis in the web is thought to be oriented perpendicular to the ND, but the observed short-range fiber axis was not oriented that much.

The fiber orientation distributions of the needle-punched samples are displayed in Figure 9. Even at a penetration depth of 6.4 mm, the probability of ND-oriented fibers was obviously larger than that of the web. The ND-oriented fibers should be produced by needle punching. In addition, for penetration depths of 12.7–19.0 mm, the probability of fibers being oriented perpendicular to the ND was also increased compared with that of the web. This increase can be explained by the compression of the fabric as follows.

Figure 9.

Fiber orientation distribution (θ) of needle-punched samples. P and D denote the needle penetration depth (mm) and needling density (punches/cm²), respectively.

The effect of the compression ratio of the fabric on the fiber orientation distribution was investigated both theoretically and experimentally. Initially, the change in fiber orientation distribution was estimated theoretically. The orientation angle of each fiber segment should be increased by web compression. If the homogeneous one-directional compression can be assumed as shown in Figure 10, the fiber orientation angle can be estimated from equation (5)

\begin{matrix} R^{2} = z^{2} + x_{0}^{2} = \frac{1}{λ^{2}} z_{0}^{2} + R_{0}^{2} - z_{0}^{2} \\ {cos}^{2} θ = \frac{z^{2}}{R^{2}} = \frac{\frac{1}{λ^{2}} {cos}^{2} θ_{0}}{(\frac{1}{λ^{2}} - 1) {cos}^{2} θ_{0} + 1} \end{matrix}

(5)

where θ₀ and θ are the inclination angles of fibers before and after compression, and λ indicates the compression ratio. The inclination angles of fibers in the original web were used for θ₀. The compression ratio λ used for the calculation was 2.24, which is almost the same as the maximum compression ratio of the needle-punched samples as listed in Table 2. In addition, the change in fiber orientation distribution was also measured experimentally by the XCT measurement for a compressed web. The compressed web was prepared by compressing the original web between a pair of acrylic plates up to a compression ratio of 2.24.

Figure 10.

Effect of compression ratio on the fiber orientation angle (θ) assuming homogeneous one-directional compression. z₀ and λ indicate the web thickness and compression ratio, respectively.

The obtained results are presented in Figure 8. The probability of fibers being oriented perpendicular to the ND clearly increased, whereas the probability of ND-oriented fibers decreased with compression for both the theoretical and experimental results. However, the probability of cosθ < 0.2 for the result of equation (5) is clearly larger than that of the compressed web. This can be explained by the compression buckling of fibers in the compressed web. In a buckled fiber, part of the fiber is oriented to the ND instead of lying along the fabric plane. The effects of needle-punching on fabric structure were examined by comparing the fiber orientation distributions of a compressed web and needle-punched sample with similar thickness, a penetration depth of 19.0 mm, and needling density of 80 punches/cm², as shown in Figure 9(c). The probability of ND-oriented fibers for the needle-punched sample was larger than that of the compressed web.

Considering the results presented in this section, schematic illustrations of the needle-punched nonwoven fabrics obtained for each penetration depth are shown in Figure 11. The obvious increase of ND-oriented fibers (cosθ > 0.7) induced by needle punching should be caused by the punching of fibers into the fabric by the needle. For penetration depths of 12.7–19.0 mm, the needle-punched samples were markedly compressed, as mentioned above. In this case, the ND-oriented fibers are thought to form a pillar structure, which holds the compression force applied to the fabric. The fabric is compressed by combination of the restrained bridging fibers and stitch structure connecting the pillars. In contrast, the probability of ND-oriented fibers for the sample with a penetration depth of 6.4 mm was larger than that for the web, but there was little compression of the fabric. This can be explained by the formation of a loop structure, which could not hold the large compression force. Unfortunately, because only the structure at the half thickness of the fabric was observed in this study, we could not distinguish between the pillar and loop structures in the obtained orientation distributions. To better understand the fabric structure, the structure distribution along the thickness direction of the fabric is required. That is, we will be able to differentiate the pillar and loop structures by determining the fiber orientation distribution across the thickness of the fabric.

Figure 11.

Model structure of needle-punched nonwoven fabrics at different penetration depths (P, in mm). z₀ and λ indicate the web thickness and compression ratio, respectively. ND: normal direction.

Three-dimensional fiber orientation

The 3D fiber orientation in the fabric samples was evaluated by the second-order orientation factor f (Herman’s orientation factor) to the reference axis.²¹ The inclination angle θ and orientation angle (ϕ) from the MD in the fabric plane were used for this calculation; here ϕ is the angle between the axis along a of an object and the MD, as depicted in Figure 7. The degree of orientation for each reference axis can be obtained by the following equations

\begin{matrix} f_{MD} = \frac{3 {sin}^{2} θ {cos}^{2} ϕ - 1}{2}, f_{CD} = \frac{3 {sin}^{2} θ {sin}^{2} ϕ - 1}{2}, \\ f_{ND} = \frac{3 {cos}^{2} θ - 1}{2} \end{matrix}

(6)

where the angle brackets 〈〉 indicate the volume average of all objects, which can be calculated by following equations using θ and ϕ of each object (i)

\begin{matrix} {sin}^{2} θ {cos}^{2} ϕ = \frac{\oint_{l} {sin}^{2} θ {cos}^{2} ϕ dl}{\oint_{l} dl} = \frac{\sum_{i} {sin}^{2} θ_{i} {cos}^{2} ϕ_{i} \frac{1}{cos θ_{i}}}{\sum_{i} \frac{1}{cos θ_{i}}} \\ {sin}^{2} θ {sin}^{2} ϕ = \frac{\oint_{l} {sin}^{2} θ {sin}^{2} ϕ dl}{\oint_{l} dl} = \frac{\sum_{i} {sin}^{2} θ_{i} {sin}^{2} ϕ_{i} \frac{1}{cos θ_{i}}}{\sum_{i} \frac{1}{cos θ_{i}}} \\ {cos}^{2} θ = \frac{\oint_{l} {cos}^{2} θ dl}{\oint_{l} dl} = \frac{\sum_{i} cos θ_{i}}{\sum_{i} \frac{1}{cos θ_{i}}} \end{matrix}

(7)

It is worth noting that, as shown in equation (6), the calculated values are not the number average but the average along the length of fiber included in the scanned ROI. Here, the sum of the orientation factors f_MD + f_CD + f_ND should always be zero, and f values of 1, 0, and−0.5 indicate the perfect axial orientation, random orientation, and perfect in-plane orientation normal to the reference axis, respectively.

The obtained orientation factors are listed in Table 2. Fibers in the original web were oriented somewhat perpendicular to the ND. However, they showed almost no orientation in the plane because the difference between f_MD and f_CD was small. The orientation perpendicular to the ND was enhanced by the web compression, whereas it was decreased by needle punching because f_ND increased by needle punching. f_ND decreased slightly from a 6.4 to a 12.7 mm penetration depth, but f_ND did not change with further penetration. As shown in the previous section, the orientation distribution was changed by the penetration depth and needling density. No clear change in f_ND resulted, despite the change in orientation distribution that can be explained by a cancelling of two effects, namely, an increase in ND-oriented fibers by the needling and the falling of fibers by fabric compression.

The in-plane orientation of fibers can be discussed by considering the difference in f_MD and f_CD. Somewhat higher f_MD than f_CD was observed for the samples with a penetration depth of 12.7–19.0 mm, while no obvious difference was observed for those with a penetration depth of 6.4 mm and the web. Because a parallel-laid web was used, its fibers should be roughly oriented to the MD rather than the CD. However, no obvious difference between f_MD and f_CD was observed, particularly for the fabric with a short needle penetration depth and the web. This slight difference is probably caused by fiber crimping, as discussed above. In contrast, for the samples with penetration depths of 12.7–19 mm, differences in fiber orientation were observed. This indicates that straightened bridging fibers between the pillars of ND-oriented fibers were formed by the deep penetration of needles.

Relationship between structural parameters and tensile strength of fabrics

Because a needle-punched nonwoven fabric consists of short fibers without any bonding, the frictional forces between the fibers are needed to impart tensile strength to the fabric. The frictional force transfers the applied external force from fiber to fiber at their contact points in the fabric. Therefore, the fiber-volume fraction in the fabric should affect its strength because the increase of the fiber-volume fraction will raise the number of contact points. In addition, because the fiber bears the tensile force applied along its length, the fiber orientation in the fabric should also affect its tensile strength.²² The probability of ND-oriented fibers in the fabric was increased by needle punching. Thus, we focused on the probability of ND-oriented fibers as a parameter illustrating fiber orientation related to the tensile force tolerated by the fabric.

Figure 12 shows the tensile strength of fabric plotted against the volume fraction of ND-oriented fibers, which is the product of the volume fraction of fibers in the fabric, and the probability of ND-oriented fibers. The volume fraction of fibers whose cosθ exceeded 0.7, corresponding to a fiber inclination angle of less than 45°, was selected as the ND-oriented fibers because their content increased following needle punching. Linear relationships between tensile strength and the volume fraction of ND-oriented fibers were observed. This indicates that the tensile strength of needle-punched fabrics can be designed quantitatively by the two structural parameters of fiber orientation and the volume fraction. However, there is a clear difference in the relationships between the strengths along the MD and CD; that is, a lower horizontal-axis intercept and steeper slope of the strength were observed in the MD than in the CD.

Figure 12.

Relationship between the volume fraction of normal direction (ND)-oriented fibers and tensile strength of the fabric. The volume fraction of ND-oriented fibers was calculated by multiplying the probability of fibers with cosθ above 0.7 by the volume fraction of fibers in a corresponding tomographic image. CD: cross-machine direction; MD: machine direction.

It is interesting that the fabric strength is decided by the amount of ND-oriented fibers because the ND-oriented fibers, which are aligned perpendicular to the tensile force, cannot directly bear the tensile force applied to the fabric. Instead, this observation can be explained by the horizontal-axis intercept. The horizontal-axis intercept suggests that a certain amount of ND-oriented fibers is required to form the force-bearing structure, either the bridging fiber or stitch structure. The higher intercept for the CD than the MD indicates that more ND-oriented fibers are required to form the CD-oriented bridged fiber than in the case of the MD. The linear relationship observed above the intercept indicates that the force-bearing structures are formed in proportion to the amount of ND-oriented fibers. The larger slope for the MD than the CD suggests that a larger amount of MD-oriented bridging fibers or stitch structures tend to be produced with the formation of ND-oriented fibers of pillar or loop structures. This can be explained by the preferential fiber orientation to the MD rather than the CD in the original web. Because the interval of needling points along the MD and CD are almost the same in this study, and more MD-oriented fibers than CD-oriented ones that are originally laid in the web, more MD-oriented bridging fibers or even stitch structures should be formed along the MD rather than the CD.

An obvious tensile strength increase caused by the insertion of ND-oriented fibers was also reported by Hearle and co-workers.^1,9 They also pointed out that the amount of ND-oriented fibers should be related to the amount of fibers bearing the tensile force. However, they stopped at the qualitative level because they had no way to analyze the structure quantitatively. The results presented here are the first to evaluate the tensile strength of needle-punched nonwoven fabrics quantitatively by considering their structural parameters. There must be other relationships between other properties and structure parameters of nonwoven fabrics; their quantitative relationships can be used to design the properties of these materials.

Conclusions

In this study, the structure of needle-punched nonwoven fabrics was analyzed by XCT, and the relationship between the needle-punching conditions and structural parameters, such as volume fraction and fiber orientation, was investigated. Moreover, the relationship between the obtained structural parameters and tensile strength was investigated. The obtained results are summarized below.

The volume fraction of fibers in the middle layer of fabric was clearly larger than that obtained by bulk measurements above the compression ratio of 1.4, while there was no clear difference below this compression ratio.

The fibers tended to be oriented to the ND of the fabric following needling. Increasing the needle penetration depth caused the fibers to become oriented perpendicular to the ND because of the compression of the fabric.

Somewhat higher f_MD than f_CD was observed for the samples with a penetration depth of 12.7–19.0 mm (from the fourth to the seventh barb reached the bottom surface of the web), while no obvious difference was observed for the web and samples with a penetration depth of 6.4 mm (the first barb reached the bottom surface of the web).

There were clear linear relationships between the tensile strength and volume fraction of ND-oriented fibers. However, the slope and horizontal intercept of these lines were different for the strengths in the MD and CD.

The obtained results demonstrate that XCT image analysis is an effective technique to quantitatively evaluate the structure of needle-punched nonwoven fabrics. It was revealed that the tensile strength of needle-punched nonwoven fabrics can be designed by controlling their structural parameters.

Footnotes

Acknowledgements

The authors thank Kanai Juyo Kogyo Co., Ltd (Osaka, Japan) for kindly supplying the fibers. We thank Laura Kuhar, PhD, from Edanz Group () for editing parts of a draft of this manuscript.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by a Grant-in-Aid for the Shinshu University Advanced Leading Graduate Program by the Ministry of Education, Culture, Sports, Science and Technology (MEXT), Japan.

References

Hearle

JWS

Sultan

MAI

Choudhari

. A study of needled fabrics Part II: effects of the needling process. J Text Inst 1968; 59: 103–116.

Hearle

JWS

Purdy

. The structure of needle punched fabric. Fibre Sci Technol 1971; 4: 81–100.

Hearle

JWS

Choudhari

. 35–A study of needled fabrics Part VII: the transfer of fibres through the web by needling. J Text Inst 1969; 60: 478–496.

Miao

. An experimental study of the needled nonwoven process Part II: fiber transport by barbed needles. Text Res J 2004; 74: 394–398.

Ventura

Ardanuy

Capdevila

et al.

Effects of needling parameters on some structural and physico-mechanical properties of needle-punched nonwovens. J Text Inst 2014; 105: 1065–1075.

Midha

Mukhopadyay

. Bulk and physical properties of needle-punched nonwoven fabrics. Indian J Fibre Text Res 2005; 30: 218–229.

Hearle

JWS

Sultan

MAI

. A study of needled fabrics Part VI: the measurement of punching force during needling. J Text Inst 1968; 59: 237–242.

Watanabe

Miwa

Yokoi

et al.

Predicting the penetrating force and number of fibers caught by a needle barb in needle punching. Text Res J 2004; 74: 417–425.

Hearle

JWS

Purdy

. On the nature of deformation of needled fabrics. Fibre Sci Technol 1972; 5: 113–128.

10.

Hearle

JWS

Purdy

. On resistance to slippage in needled fabrics. Fibre Sci Technol 1978; 11: 127–144.

11.

Gilmore T, Davis H and Mi Z. Tomographic approaches to nonwovens structure definition. National Textile Center Annual Report, USA, September 1994.

12.

Manickam

McCutcheon

. Characterization of polymeric nonwovens using porosimetry, porometry and X-ray computed tomography. J Memb Sci 2012; 407–408: 108–115.

13.

Soltani

Johari

Zarrebini

. Effect of 3D fiber orientation on permeability of realistic fibrous porous networks. Powder Technol 2014; 254: 44–56.

14.

Tausif

Duffy

Grishanov

et al.

Three-dimensional fiber segment orientation distribution using X-ray microtomography. Microsc Microanal 2014; 20: 1294–1303.

15.

Chen

Koker

MKA

Uzun

et al.

In-situ X-ray study of the deformation mechanisms of non-woven polypropylene. Int J Solid Struct 2016; 97–98: 200–208.

16.

Jeon

Kim

et al.

Predicting the tensile strength of needle-punched nonwoven mats using X-ray computed tomography and a statistical model. Fibers Polym 2014; 15: 1202–1210.

17.

Jeon

Choi

et al.

In situ monitoring of structural changes in nonwoven mats under tensile loading using X-ray computer tomography. Composites Part A 2014; 63: 1–9.

18.

Ishikawa

Kim

Oshkosh

. Visualization of a pillar-shaped fiber bundle in a model needle-punched nonwoven fabric using X-ray micro-computed tomography. Text Res J 2017; 87: 1387–1393.

19.

Russell

. Handbook of nonwovens, 1st ed. Cambridge: Woodhead Publishing Ltd, 2006, pp. 227–227.

20.

Bernasconi

Cosmi

Hine

. Analysis of fibre orientation distribution in short fibre reinforced polymers: a comparison between optical and tomographic methods. Compos Sci Technol 2012; 72: 2002–2008.

21.

Herman

. Physics and chemistry of cellulose fibers, Amsterdam Newfork: Elsevier, 1949, pp. 247–247.

22.

Komori

Makishima

. Numbers of fiber-to-fiber contacts in general fiber assemblies. Text Res J 1977; 47: 13–17.