Design and characterization of a bicomponent melt-spun fiber optimized for artificial turf applications

Full Paper

Design and Characterization of a BicomponentMelt-Spun Fiber Optimized for ArtificialTurf Applications

Rudolf Hufenus,* Christian Affolter, Martin Camenzind, Felix A. Reifler

Artificial turf is robust, playable in all weathers and has a long service life. Polyamide (PA)flooring has excellent resilience but provokes abrasion injuries (friction burn); polyethylene(PE) monofilaments are skin-friendly but tend to permanent deformation. To maximizeresilience while minimizing the risk of skinabrasion, PA-PE bicomponent fibers are devel-oped. Numeric simulation is applied to findoptimized fiber cross-sections and material com-binations, accompanied by melt-spinning ofrespective filaments and validation of the model.The resulting artificial grass resembles naturalturf with respect to playability and appearanceand does not need any granular infill.

1. Introduction

Access to a sports field which is playable all year round is

highly requested by soccer players. The use of artificial turf

allows to play throughout the year under any weather

conditions with much less maintenance.[1,2] The world’s

first artificial sports pitch was installed in the year 1966 in

the Houston Astrodome.[3] However, it was found that

surfaces with synthetic grass produced more abrasion

injuries (friction burn) than natural grass.[4] A common

R. Hufenus, F. A. ReiflerLaboratory for Advanced Fibers, Empa, Swiss Federal Laboratoriesfor Materials Science and Technology, Lerchenfeldstrasse 5, CH-9014 St. Gallen, SwitzerlandE-mail: [email protected]. AffolterLaboratory for Mechanical Systems Engineering, Empa, SwissFederal Laboratories for Materials Science and Technology,Ueberlandstrasse 129, CH-8600 Dubendorf, SwitzerlandM. CamenzindLaboratory for Protection and Physiology, Empa, Swiss FederalLaboratories for Materials Science and Technology,Lerchenfeldstrasse 5, CH-9014 St. Gallen, Switzerland

Macromol. Mater. Eng. 2013, 298, 653–663

� 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim wileyonlin

complaint about first and second generation synthetic turf

was its tendency to produce friction burns when a player,

e.g., made a sliding tackle on the surface.[5,6] This problem

could be overcome in the late 1990s by introducing so-called

third-generation artificial turf.[7] Here, the former poly-

amide (PA) yarn is replaced by polyolefin monofilaments

supported by rubber granules.[8] Polyethylene (PE) and

especially linear-low density polyethylene (LLDPE) has

become the raw material of choice for synthetic grass

blades, offering reduced skin abrasion and superior player

friendliness in sliding and tackling compared to other yarn

raw materials.[9]

The third-generation artificial turf comprises a playing

surface (pile) made of synthetic grass yarn, a support

backing on which the yarn is sewn, and an infill to improve

the rebound resilience of the playing surface.[10] The

widespread use of rubber infill, obtained by grinding used

tires, is believed to cause environmental risks.[11]

Recycled crumb rubber contains a number of chemicals

that are known or suspected to cause health effects.[12–14]

Contact with recycled tire crumb infill is also known to

reduce the efficacy of UV stabilizers used in PE grass yarn.[15]

Thus, and due to economic reasons (erosion of infill

elibrary.com DOI: 10.1002/mame.201200088 653

654

www.mme-journal.de

R. Hufenus, C. Affolter, M. Camenzind, F. A. Reifler

material), artificial turf without infill is more favorable.

A problem that has to be addressed when omitting the

supporting material is that polyolefin monofilaments tend

to permanent deformation (creep).[16] This means that after

being used for some time the pile becomes completely

flattened, a condition which is not only optically unplea-

sant but also results in changes of the playing properties of

the surface. While the ability of polyolefin fibers to recover

from a deformation is poor, PA fibers are known for their

excellent resilience.[17]

However, as mentioned above, PA fibers – in contrast to

polyolefins – are attested to provoke abrasion injuries.[18,19]

To our knowledge this practical experience is still lacking

fundamental prove, on one hand because medical reports

about abrasion injuries on artificial turf do not differentiate

between types of synthetic grass, on the other hand

because studies about friction of human skin against textile

materials concentrate on low speed.[20] A possible explana-

tion of the experienced skin-friendliness of PE under severe

sliding conditions is that the softening temperature of the

polymer can be reached during sliding.[21] The softened

polymer forms a low shear-strength interfacial layer at the

sliding surface, which behaves as a lubricant.[22] As the

softening point of PA is considerably higher than that of

PE,[23] higher friction temperatures and thus higher risk of

skin injuries can be expected for PA fibers.

Bicomponent fibers are among the most interesting

developments in the field of synthetic fibers.[24–26] They are

synthetic fibers made from two polymers of different

chemical and/or physical structure, extruded from a

common spinneret to form a single filament.[27] The

polymer flows are kept separate up to the spin pack and

brought together in or before the spinneret capillary. When

the filament leaves the spinneret, it consists of non-mixed

components that touch at the interface. Depending on the

characteristics of the different polymers, bicomponent

fibers can act as bonding elements in thermobonded

nonwoven fabrics, as self-crimping fibers to achieve

textured yarn, or as fibers with the surface functionality

of special polymers and additives at reduced cost.[25,28] The

core/sheath approach enables a variety of surfaces while

maintaining major fiber and textile properties, because

most thermoplastic polymers can be applied as a sheath

over a core that provides the requested tensile strength.[29]

Bicomponent fibers pass through common melt-drawing

processes similar to conventional synthetic fibers.[26] If they

contain flexible-chain polymers like PA or PE, they can only

be partially oriented in the fluid state during the melt-

spinning step. Their orientation has to be completed in the

solid state, i.e., in the post-spinning drawing step, and for

such fibers, drawing is the principal mean of building up

their tensile properties. The orientation is a function of

the actual strain, and a high degree of orientation can be

produced without application of high stress if the strain rate

Macromol. Mater. Eng. 2

� 2013 WILEY-VCH Verlag Gmb

is low. During the drawing process, the as-spun filament

is irreversibly stretched up to several times its original

length. The elongation is accompanied by extension

and parallelization of macromolecules and crystallites

along the fiber axis. Increasing both the orientation and

the degree of crystallinity through the drawing process

finally leads to a fiber with greatly improved tenacity and

modulus.[30–32]

A bicomponent fiber with a PA core and a PE sheath is

expected to render a robust, skin-friendly artificial turf. The

goal of this work was to develop a respective monofilament

with optimized cross-section and material combination to

maximize resilience while minimizing risk of skin abrasion,

in order to achieve artificial grass for sports flooring that

resembles natural turf.

2. Experimental Section

2.1. Materials

Dowlex SC 2108, an LLDPE, was provided by Dow Plastics (Dow

Europe GmbH, Horgen, Switzerland). The density was

0.935 g � cm�3, the melting temperature Tm was 250 8C, and the

melt flow index (190 8C, 2.16 kg) was 2.5 g � (10 min)�1. As prepara-

tion for melt spinning, the LLDPE was dried in a hot-air cabinet at

50 8C for 12 h.

Grilon F34, a polyamide 6 (PA6), was provided by Ems-Grivory

(Ems-Chemie AG, Domat/Ems, Switzerland). The density was

1.14 g � cm�3, the melting temperature Tm was 222 8C, and the melt

flow index (275 8C, 5 kg) was 35 g � (10 min)�1. As preparation for

melt spinning, the PA6 was dried in a vacuum cabinet at 100 8Cfor 12 h.

The synthetic grass monofilaments assessed for reference

purposes were taken form commercially available artificial turf

yarns. To evaluate the material properties used as input data for

numeric simulation, monofilaments from an 8�750 dtex PA6

(PA750) yarn provided by Schramm (Rahden, Germany) and a

2�6900 dtex PE (PE6900) yarn provided by Tisca Tiara (Buhler and

Urnasch, Switzerland) were considered. For comparative purposes

an 8�1750 dtex PA6 (PA1750) yarn extracted from a customary

artificial turf (Domo Champion Infinity, Domo, Sint-Niklaas,

Belgium) and the 8� 1550 dtex PE (PE1550) yarn Diamond LSR

MF (Ten Cate, Nijverdal, The Netherlands) were selected.

2.2. Melt-Spinning Equipment

To assist and validate the numeric simulation performed in this

study, melt-spinning of model fibers was carried out on Empa’s

custom-made pilot melt-spinning plant built by Fourne Poly-

mertechnik (Alfter-Impekoven, Germany). This plant, with features

corresponding to an industrial plant, enables the production of

mono- and bicomponent fibers with various fiber cross-sections

and material combinations with a throughput of 0.1–5 kg �h�1. It

comprises two screw extruders and one piston extruder, a spin pack

with thermally discrete polymer conduits, and a set of spinnerets

allowing for elaborate fiber cross-sections. The pilot melt-spinning

plant is further specified in an earlier paper.[33]

013, 298, 653–663

H & Co. KGaA, Weinheim www.MaterialsViews.com

Table 1. Extrusion parameters and physical properties of the model fibers produced. Stated are the composition of the core-sheath fibers,the temperatures of the polymers leaving the extruder barrel, as well as dimensional and tensile properties measured on the as-spun fibers.All fibers were produced with a draw ratio of DR¼4. The linear mass density was measured by weighing 100 m of the fibers sampled underdefined strain. Fiber and core diameters were calculated taking linear mass density, density of the two polymers, and percentage of core andsheath into account.

Fiber

no.

Polymer Composition

[wt%]

Temperature

[-C]

Diameter

[mm]

Linear

mass

density

[dtex]

Tensile

strength

[cN/tex]

Ultimate

tensile

stress

[GPa]

Ultimate

tensile

strain

[%]Core Sheath Core Sheath Core Sheath Core Total

451 PA – 100 – 260 – – 77 53 51� 5 0.58� 0.06 37� 7

452 PA PE 74 26 260 245 76 88 66 46� 5 0.50� 0.06 47� 9

453 PA PE 59 41 260 245 77 100 83 41� 1 0.43� 0.01 52� 6

454 PA PE 50 50 260 245 76 108 95 35� 2 0.37� 0.02 47� 5

455 PE – 100 245 – 67 40 24� 2 0.22� 0.02 107� 7

457 PE – 100 245 – 79 56 18� 3 0.17� 0.02 96� 26

Design and Characterization of a Bicomponent . . .

www.mme-journal.de

For this study, the two polymers were melted using

two single screw extruders, and coaxially combined in a

spinneret for bicomponent monofilaments with core/sheath-

geometry. The extruder temperatures were 260 8C for the PA

used, and 245 8C for the PE, respectively (Table 1). The spin

pack temperature was 260 8C for all fibers. The die consisted of

a tube with 0.4 mm inner diameter and 0.7 mm outer

diameter within a 1.2 mm capillary. The draw ratio (ratio of speed

of draw and feed godets) was set to 4, the winding speed to

1200 m �min�1.

Coordinated by Schramm (Rahden, Germany), the prototype

artificial turf fiber (PATF) was produced on a bicomponent

monofilament line provided by Reimotec (Ober-Abtsteinach,

Germany). The respective spinneret was built by HEH (Erlensee,

Germany).

2.3. Fiber Characterization

The surface morphology of fibers was analyzed using the Hitachi

FE-SEM S-4800 scanning electron microscope (SEM, Hitachi High-

Technologies Europe, Krefeld, Germany).

The load-strain behavior of small scale fibers melt-spun on

Empa’s pilot plant was evaluated using the Tensorapid 3 tensile

tester (Uster Technologies, Uster, Switzerland); 500 N load cell;

single filament tests with 100 mm test length and a constant rate

of extension of 100 mm �min�1. The same apparatus and test

conditions were used to perform cyclic loading tests on the fibers

considered. The fibers were extended to an elongation (strain level)

of 5, 10, 15, and 20%, respectively. For each preset elongation,

a virgin fiber was repeatedly extended and released at a constant

rate of 100 mm �min�1 to achieve five loading cycles.

Creep tests were performed in a frame enabling

deformation measurements, using dead-weight loadings of

20 and 40% of the respective maximum tensile strengths of the

fibers tested.

A Lisport wear tester (Labosport, Le Mans, France) was operated

by Tisca Tiara (Buhler and Urnasch, Switzerland) to assess the

durability of an artificial turf.[9]

www.MaterialsViews.com



2.4. Numeric Simulation

The finite element (FE) solver used for this project was Abaqus/

Standard Versions 6.6-EF1 to 6.8-EF1.[34] All simulations were

performed nonlinear because large strains and deformations had to

be considered, and due to the implemented material models.

Geometric contact between fibers and rigid parts (for time

dependent loading of the fibers) was modeled with the ‘‘augmented

Lagrange’’ approach in Abaqus, friction was neglected (m¼ 0). The

material models considered hyperelasticity for the short-term

response of the fibers, and viscoelasticity for consideration of time

dependent effects such as creep or stress relaxation. Plasticity or

visco-plasticity was not considered. The fibers were modeled with

linear ‘‘hybrid’’ brick elements in Abaqus (C3D8RH or C3D8H),

which are adequate for quasi incompressible and/or hyperelastic

materials.

For the assessment of different fiber cross-sections concerning

their mechanical performance (elastic restoring force, resilience

after creep), a comparative FE study on a single fiber model was

performed. The geometric model with its boundary conditions

(BCs) for the numerical investigation of a single fiber is shown in

Figure 1 (left).

The fiber with an initial curvature (R¼ 50 mm) is fully

constrained on the ground and has a total height of 20 mm. A

solid plate (‘‘analytical rigid’’ in Abaqus) is rotated from an initially

quasi vertical position into a position parallel to the ground (final

distance 5 mm). The fiber thus experiences an increasing bending

load over the resulting contact point to the plate.

The loading happens in a relatively short time of 2 s, cf. step 1 in

Figure 1 (right). The dwell (time in which the stress relaxation

during the external loading takes place) was set to 600 s. The

continuous line indicates the applied deformation (fixed BC), and

the model had to predict the resulting elastic restoring force in step

1 (dashed line).

After the relaxation, the plate was lifted again in 2 s. Due to the

creep strains in PA or PE, the fiber could not follow the plate

anymore. The resulting tip displacement of the fibers vs. time was

identified in step 2 (dotted line) and then compared for the various

fiber cross-sections.

013, 298, 653–663

H & Co. KGaA, Weinheim655

https://www.researchgate.net/publication/237099903_Combined_testing_for_MEMS_characterization?el=1_x_8&enrichId=rgreq-e6218a3f-f31b-486e-9839-822ee1878d95&enrichSource=Y292ZXJQYWdlOzI1NTgyNTY2NDtBUzoxOTY0MTQxMTA5OTg1NDFAMTQyMzg0MDE4MTU2NA==

Figure 1. Geometric model with applied BCs at t¼0 s (step 0), and t¼ 2 s (step 1), respectively (left), as well as loading and unloading of thefiber (displacement control), expected restoring force and resulting tip displacement at unloading (time dependent; step 2) (right). Thedotted lines show the general shape of the respective curves; examples of calculated curves are presented in Figure 7.

656

www.mme-journal.de


2.5. Stress Relaxation and Elastic Recovery Tests

To verify the predictions of the material models developed for the

FE analyses in this study, a 2-point bending method developed at

the Centre of Mechanics (IfM) at ETH (Zurich, Switzerland) was

implemented to monitor the restoring forces over time with a

highly sensitive load cell.[35] A cut segment of a fiber was positioned

manually into the test setup, which consisted of two parallel

supports with a milled groove on each side (Figure 2). Care had to be

taken to place the fibers in a perfect horizontal plane, preventing

the fiber from bouncing out. After visual inspection of the correct

orientation the two supports were closed with a controlled speed

of v¼0.5 mm � s�1. The final gap width was 1.0 mm, and in this

position the resulting horizontal force was measured over �5 min

(stress relaxation).

A method to assess elastic recovery with a high-speed camera

MotionXtra HG-100K (Redlake, San Diego, USA) was implemented

Figure 2. Schematic setup of the stress relaxation test (left) and top



in order to validate the numeric model (simulate buckling of a fiber

when an artificial turf is played on). For this purpose, one end of the

fiber was fixed in an epoxy embedding, and the other end was

pressed down by a plunger mounted and moved perpendicular to

the fiber stem, as illustrated in Figure 3 (left). The model fibers spun

on Empa’s pilot plant were chosen to protrude 6 mm, and to be bent

down to a chuck gap of 1 mm. In the case of the thicker synthetic

grass monofilaments, the protrusion was 20 mm and the gap 5 mm.

To conduct a guided distortion, a groove was added to the movable

plunger. The setup was coupled with a static testing machine, in

order to monitor the relative displacement of fiber mounting and

depressing plunger. The fiber bending was recorded with a rate of

500 frames � s�1, the pictures were analyzed and the angle between

embedding plate and base-to-tip line of the fiber was plotted

versus time after removal of the depressing plunger. That way,

complete recovery would be represented by an angle a of 908(Figure 3, right).

view of a strained fiber (right).

013, 298, 653–663


Figure 3. Schematic setup of the elastic recovery test (left) and assessment of the recovery angle (right). The dimensions of the test setupwere adapted to the fibers (model fibers or synthetic grass monofilaments, respectively).


www.mme-journal.de

3. Results and Discussion

3.1. Melt-Spinning of Model Fibers

In melt-spinning, the molten polymer emerging from the

spinneret is either quenched using a water bath or an air

current. In the pilot plant used for this study air is applied

for cooling, which is sufficient for final fiber diameters of

up to �150 mm. In contrast, artificial turf is made from

monofilaments that are too thick to be cooled by air during

spinning and thus are spun in a water bath before draw-off,

a technology requesting a setup different to the one

available at the beginning of this study. As a consequence,

the fibers spun with the pilot plant were chosen to be

approximately ten times smaller in diameter than mono-

filaments typically used for artificial turf. Nevertheless the

fibers were valuable tools to validate the numeric model.

In bicomponent melt-spinning two molten polymers

are merged after leaving the spinneret capillary, so that the

fibers consist of two joined components. To overcome the

respective drawbacks of pure PE and PA fibers, the concept

of combining the properties of PA and PE in a bicomponent

fiber was followed, and model fiber monofilaments with PA

core and PE sheath were melt-spun on Empa’s pilot plant

(Table 1). For reference purposes, additional monocompo-

nent PA and PE monofilaments were produced (fibers 451,

455, and 457). Fiber 451 is a pure PA monofilament with a

diameter of 77 mm. The bicomponent fibers 452–454 all

have a PA core with a diameter of 76 or 77 mm, fibers 455 and

457 are pure PE monofilaments. Table 1 gives a summary

of the physical properties measured on the model fibers

produced.

To assure comparability, the draw ratio was set to 4 for all

fibers spun on the pilot plant. For the pure PE fibers (455 and

457), however, microscopic cross-sections showed varying




fiber diameters of 60—100 (455) and 70–130 mm (457),

respectively. Hence, in the case of pure PE fibers,

the draw ratio was not sufficient to achieve fully drawn

fibers.

3.2. Elastic and Creep Behavior of the Model Fibers

The material parameters required for numeric simulation

had to be measured directly on extruded fibers, as they may

differ severely from values derived from semi-finished

parts (i.e., samples cut out of PA or PE blocks). Based on cyclic

loading tests (quasi-static) performed on mono-component

and bicomponent fibers with a PA core and a PE sheath, the

elastic material properties were determined according to

Figure 4: the initial loading (first pull) determines the

envelope for initial straining of the material, shown as

dashed line. Stress versus strain curves for different strain

levels of the same material were derived after cyclic loading

until steady state (i.e., after five loading-unloading cycles).

The results of the uniaxial static tensile tests on fibers are

used to describe the instantaneous elastic response of both

materials using strain energy potential functions for hyper-

elastic materials in the FE model (required for large strains

and nonlinear s/e characteristics). If experimental data for

the materials are available, the FE solver Abaqus allows the

test of different strain energy potentials in order to best fit

the measured data to the numeric model, and to get a

stable mesh deformation at large strains (i.e., finally good

convergence). The curves for a maximum strain level of 10%

were used to find the best fit:

(i) O

013,

H &

gden first order was chosen for PA with m1¼ 70.15,

a1¼ 25.0, and D1¼ 8.64� 10�4

(ii) P
olynomial with N¼ 1 (Mooney Rivlin) was used to
model PE; C10¼ 49.31, C01¼ –20.9, D1¼ 1.067� 10�3.

298, 653–663

Co. KGaA, Weinheim657

Figure 4. Elastic material properties of the PE and PA considered for numeric simulation: stress versus strain curves for cyclic loading, plusenvelope (curves with lowest and highest loads) for initial loading (single pull test).

658

www.mme-journal.de


A viscoelastic material law suitable for the implementa-

tion into an FE solver was derived. The time dependent

material properties for creep were determined from creep

tests on the melt-spun model fibers, measuring the

displacement versus time for a dead load straining the

fiber. The measured tensile creep test data were converted

into shear test data, and based on these a Prony series in

Abaqus was fitted. Material plasticity and viscoplasticity

(permanent deformation in the viscous domain) were

neglected. The material models were validated by recalcu-

lating the mechanical tests on single material fibers.

3.3. Model Validation: Stress Relaxation

Like in the numeric model, we intended to simulate the

buckling of a fiber with the same kind of loading, i.e., with

the same geometric arrangement and BCs. Figure 2 (right)

shows the top view of a tested fiber during loading. We

focused on measurements of stress relaxation during the

Figure 5. Left: restoring force versus time due to stress relaxation (sample). Right: compiled results for model fibers 451–457.



applied deformation, as the test rig only could record forces

but not strains or displacements of the fibers. The recovery

(resilience) of the fiber after unloading could not be

determined as the speed of the support was too low for a

fast unloading, and the final position of the fiber would

have been arbitrary. Thus, the measurement of the

geometric shape of the fiber versus time, which is necessary

for an assessment of the fiber resilience, was not possible.

A displacement controlled loading of the fibers according

to step 1 in Figure 1 was induced. Reaction force versus time

was measured for �5 min. The detected maximum force

Fmax at the end of the loading procedure was sensitive to

loading speed and sampling rate of the load cell.

For each material, ten samples were measured. At each

test, the maximum elastic force Fmax at the end of the

loading process (started at time t¼ 0 s), and the remaining

force 300 s after maximum force had occurred (F300), were

measured (Figure 5, left). The ratio F300/Fmax was taken as a

measure for the stress relaxation in the fiber, cf. Table 2.

showed example: model fiber 455, data for five repetitions on one

013, 298, 653–663


Table 2. Stress relaxation results for ten samples of fiber 451. Themaximum elastic force Fmax at the end of the loading and theremaining force F300 detected 300 s after the maximum force hadoccurred are listed and compared.

Test no. Fmax

[mN]

F300

[mN]

F300/Fmax

1 6.0 2.71 0.45

2 7.0 3.00 0.43

3 5.7 2.40 0.42

4 7.3 3.19 0.44

5 6.8 2.68 0.39

6 5.2 2.09 0.40

7 5.6 2.21 0.39

8 5.5 2.09 0.38

9 5.3 2.07 0.39

10 7.6 2.93 0.39

mean� st. dev. 6.20� 0.89 2.54� 0.42 0.408� 0.024


www.mme-journal.de

These values were averaged at the end and compared

(Figure 5, right). While the force values scattered consider-

ably (max. 15–20% from the mean value), the calculated

force ratios were within 8% of the mean value. In log time

scale, the decrease of restoring force after �1 min was

usually linear, compare Figure 5 (left). A comparison of the

logarithmic decrement was intended.

For a discussion of the results (Figure 5, right), the

following has to be considered: The samples from 451 to 454

contain a PA core with a diameter of 76 or 77 mm, with a PE-

sheath of increasing thickness (no sheath in the case of 451).

The samples 455 and 457 are pure PE-fibers with a diameter

of 67 and 79 mm, respectively.

A summary of the most important findings of the stress

relaxation measurements:

(i) I

www

t is obvious that fibers containing PA show a higher

initial restoration force than pure PE fibers.

(ii) F
max of the fibers containing PA increases from 6.2 to
10 mN with increasing PE sheath. The maximum

force for 455 and 457 (pure PE) is in the range of 1.6–

3.0 mN.

(iii) F
or F300/Fmax one would expect decreasing values for
increasing PE content (i.e., increasing contribution of

the viscoelastic stress relaxation in PE). However, this

was only partially true, as interaction forces between

core and sheath interfered.

(iv) A
further proof of the ability of fibers for the
(viscoelastic) recovery after loading (step 2 in

Figure 1, right) turned out to be necessary, which is

described in the following.

.MaterialsViews.com



3.4. Model Validation: Elastic Recovery

For the final application, not primarily restoring force, but

recovery from a deflection is of relevance. To assess elastic

recovery, same as in the numeric model, we intended to

simulate the buckling of a fiber. Recovery measurements

were executed with bicomponent and reference mono-

component fibers produced. In principle, the results

confirmed the findings of the numeric simulation, though

the absolute values differed considerably. Possible reasons

for these dissimilarities were differing material properties,

size effects, as well as initial deflection of the filaments due

to small intrinsic stresses.

Figure 9 (left) shows the results of the viscoelastic

recovery tests performed on the as-spun model fibers. The

pure PA fiber (451), as expected, recovered significantly

faster than the pure PE fibers (455 and 457). In comparison,

the PA-PE bicomponent fibers displayed a mean resilience

that exceeded the performance of the pure PE fibers, but

that fell behind the efficiency of the pure PA fiber. This was

true albeit the core of each bicomponent fiber had the same

dimension as the pure PA fiber. In other words, the PE sheath

retards the recovery of the PA core which was also proved

with the tests on various monofilaments (Figure 9, right),

where the bicomponent PATF recovered to the same level as

the pure PA synthetic grass monofilament, but with a

certain time lag.

3.5. Development and Selection of the New Fiber

Cross-Section

To select the optimum cross-section for the new artificial

turf fiber, nine different fiber cross-sections were developed

(Figure 6) and analyzed with the FE model described in

Figure 1. Their mechanical performance was computed and

the results were compared with respect to the instanta-

neous elastic response (max. force) as well as resilience

(long-term deformations).

The cross-sections with one circular core (1–4 and 8) are

highly symmetric (mirror symmetry, 1–3 are also axially

symmetric). Here, the focus was on simplicity for the

tool manufacturing and processing. Cross-sections 4 and

8 are based on 3: in 4, a third wing is introduced to render

a hollow Y-profile, and in 8, a cut is introduced into the

round hollow core. This cut might improve the humidity

balance, i.e., the storing and emitting of water over a

long time.

Compared to PE, PA is stiffer and less prone to creep. In

cross-sections 5–7, the goal was to maximize the contribu-

tion of PA to the bending stiffness, and it was placed as far

away as possible from the neutral axis of bending. In these

complex cross-sections with two and/or non-circular cores,

the PA is proportionately strained higher and would better

aid the recovery.

013, 298, 653–663


Figure 6. Fiber cross-sections (1–9) designed for numeric simulation purposes. The light shade represents PE, the dark shade represents PA.Hollow cores are represented by white zones.

660

www.mme-journal.de


Figure 7 (right) shows the calculated vertical fiber tip

displacement versus time at unloading for the cross-

sections shown in Figure 6. A displacement of 0 mm refers

to the initial (unloaded) state, thus the faster a fiber type

reaches this position, the faster it has recovered from the

viscoelastic deformation.

The highly symmetric cross-sections 1–3 turned out to be

the best concerning resilience. Increasing the cross-section

Figure 7. Left: calculated elastic vertical restoring force of the most relerelaxation (step 0/1) curve in Figure 1. Cross-section 4 has additionallfiber tip displacement over time at unloading (viscoelastic recovery) foFigure 1. For cross-section 7, twisting under load led to abortion of t



area mainly affects the elastic restoring force but also

improves the resilience (cf. 2 vs. 1). The calculated

displacement curve of 3 (hollow core) lies between

the curves of 1 and 2. The mechanical performance

(resilience) of 4 is only slightly lower compared to the

cross-sections 1–3.

While the highly symmetric cross-sections (1–3 and 8)

rendered promising performance under compression, the

vant cross-sections of Figure 6 during loading/relaxation, cf. loading/y been calculated with PE as sole polymer. Right: calculated verticalr the cross-sections shown in Figure 6, cf. unloading (step 2) curve inhe simulation.

013, 298, 653–663


Figure 8. Details of the constrained lower end of cross-section 4; shown is the calculatedmaximum principal strain just after the load has been applied (left; beginning of step 1,t¼ 2 s, see Figure 1) and after 600 s of relaxation (right; end of step 1, t¼602 s, seeFigure 1). Note the increasing local buckling of the innermost wing under compression.


www.mme-journal.de

less symmetric (4–7) and lentoid (9) cross-sections resulted

in undesirable twisting and local buckling. This is demon-

strated for 4 in Figure 8: its deformed state shows increasing

local buckling of the innermost wing under compression

during relaxation (step 1). For cross-section 7, twisting

under load even led to the abortion of the FE simulation

calculations.

The comparison of the two curves for cross-section 4 with

and without PA core (Figure 7, left) demonstrates that the

presence of PA in the core is an important factor for the fiber

stiffness: the elastic restoring force of the pure PE fibers is

only half of the restoring force of the fiber with PA in the

core. Cross-sections 4 with one PA core (hollow profile) and 5

with two PA cores exhibit the maximum vertical restoring

force (�0.096 N, Figure 7, left), while cross-section 9 (flat PE

lense) shows a minimum vertical restoring force of only

�0.0018 N.

From the numeric FE simulation results presented here, it

can generally be deduced that highly symmetric PA-PE

Figure 9. Viscoelastic recovery (resilience) of the model fibers (protrusion 6 mm, buckling drecovery (resilience) of the PATF (PA-PE bicomponent fiber) compared to customary artifi20 mm, buckling down to a chuck gap of 5 mm, right). Plotted is the restoring angle (908¼the depressing plate.


Macromol. Mater. Eng. 2013, 298, 653–663

� 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinh

bicomponent fibers with a circular PA core

(like in cross-sections 1 or 2) are expected

to deliver the best performance with

respect to resilience. Hollow fibers with

a tubular PA core (like in cross-sections 3

or 8) would yield a similar performance

with less material consumption. Because

of undesirable twisting and local buckling

(secondary deformations during stress

relaxation), there are major concerns

about the form stability of cross-sections

4 to 7 fibers. These undesired local

inelastic strains further delay the recov-

ery of the fibers.

For the production of the PATF, the FE

results had to be evaluated considering

additional criteria such as thermal

behavior (cooling of the fiber), wall thicknesses and

processability.

3.6. Prototype Artificial Turf Fiber (PATF)

The decision was made to start the production of a PATF

with a cross-section similar to cross-section 1 (Figure 6),

with small corrections intended to prevent fibrillation of

the sheath, such as larger wall thickness and rounded

corners (Figure 10, left). To achieve a certain adhesion

between core and sheath, maleic-anhydride-grafted poly-

ethylene (PE-g-MAH) was added to the PE fraction, a method

commonly used to induce compatibility between PAs

and PE.[36]

Recovery measurements were executed on the prototype

artificial turf PA-PE bicomponent fiber and compared to

reference synthetic grass monofilaments (Figure 9, right).

In principle, the results confirmed the findings of the

numeric simulations, though the absolute values differed

own to a chuck gap of 1 mm, left) and viscoelasticcial grass PA and PE monofilaments (protrusioncomplete recovery) versus time after removal of

eim661

Figure 10. SEM cross-section of the PA-PE bicomponent PATF (left), and microscopic image of the artificial turf made from this fiber, afterLisport wear test (right).

662

www.mme-journal.de


considerably. The commercially available PA synthetic

grass monofilament (PA1750), as expected, recovered

completely and significantly faster than the PE synthetic

grass monofilament (PE1550). In comparison, the resilience

of the prototype artificial turf PA-PE bicomponent fiber

is clearly better than that of the pure PE synthetic grass

monofilament. Initially, the bicomponent fiber rebounds

slower than the pure PA monofilament, but after �1 s

its recovery rate gains the lead (Figure 9, right).

The PATF was used by Tisca Tiara (Buhler and Urnasch,

Switzerland) to produce an artificial turf. Unexpectedly this

artificial turf failed the Lisport wear test; the sheath

fibrillated and detached itself from the core (Figure 10,

right). Subsequently, the industry partners modified

the fiber cross-section experimentally and finally came

up with a bicomponent fiber that comprises several PA

cores in a PE sheath. The resulting artificial turf has been

installed on several sports fields and has proven itself in

practice.[37]

4. Conclusion

We succeeded in producing bicomponent fibers for unfilled

artificial turf that show better resilience than up-to-date

monocomponent synthetic grass, without cut-back in skin-

friendliness. This goal was achieved by developing PA-PE

core-sheath monofilaments with numerically predicted

optimal cross-sections to maximize resilience while mini-

mizing risk of skin abrasion. The nonlinear FE simulation

allowed to predict the mechanical long-term behavior of

new cross-sections and material combinations and reduced

the need for extensive prototype testing. Complex fiber

cross-sections with multiple cores and asymmetric cross-



sections could be rejected at an early stage, because of their

risk for local buckling and twisting (additional creep strains

which reduce resilience).

Though, to some extent, the viscoelastic behavior of the

PE sheath hinders the recovery of the PA core, the softness of

the resulting artificial turf considerably increases the

comfort for the player. The fiber developed represents

one step further in copying natural grass. The performance

of the now commercialized artificial turf based on this fiber

is promising. A main draw-back remains: surface tempera-

tures of synthetic turf are significantly higher than natural

grass surfaces when exposed to sunlight,[38] which can

contribute to physiological stress of athletes and can cause

heat-related illnesses.[39] This is still an unsolved problem

that requires further research efforts.

Acknowledgements: This research was co-funded through a grantby the Swiss Innovation Promotion Agency CTI and through acontribution from Tisca Tiara (Buhler and Urnasch, Switzerland).The authors thank Benno Wust for operating the spinning plant,Pierluigi Barbadoro and Rolf Stampfli for mechanical character-ization, Patrick Rupper, Marcel Halbeisen and Rahel Vetter formicroscopic characterization, Adriaan Spierings and Stefan Buobfrom Inspire (St. Gallen, Switzerland) for spin pack design andproduction, Andreas Schifferle from IfM at ETH (Zurich, Switzer-land) for relaxation tests, Gerhard Schramm from Schramm GmbH(Rahden, Germany) for his valuable input and for monofilamentproduction, as well as Andreas Tischhauser from Tisca (Buhler,Switzerland) and Kaspar Zogg from Tiara (Urnasch, Switzerland)for their contribution in initiating and completing this research.

Received: March 13, 2012; Revised: May 22, 2012; Publishedonline: August 24, 2012; DOI: 10.1002/mame.201200088

013, 298, 653–663



www.mme-journal.de

Keywords: artificial turf; bicomponent fibers; modeling; poly-amides; polyethylene

[1] A. McLeod, ‘‘The Management and Maintenance of SecondGeneration Sand-filled Synthetic Sports Pitches’’, PhD thesis,Cranfield University, Cranfield 2008.

[2] V. Senner, Kunstst. Int. 2010, 100, 88.[3] J. W. McCurry, Tech. Textil. Int. 2005, 14, 11.[4] J. Ekstrand, B. M. Nigg, Sports Med. 1989, 8, 56.[5] FIFA. Quality Concept for Artificial Turf, Federation Interna-

tionale de Football Association, Zurich 2006.[6] J. L. Dragoo, H. J. Braun, Sports Med. 2010, 40, 981.[7] J. Ekstrand, T. Timpka, M. Hagglund, Br. J. Sports Med. 2006, 40,

975.[8] S. Waterman, Panstadia Int. 2003, 9, 32.[9] P. Sandkuehler, E. Torres, T. Allgeuer, Procedia Eng. 2010, 2,

3367.[10] G. Fisher, Tech. Textil. Int. 2005, 14, 11.[11] J. Zhang, I. K. Han, L. Zhang, W. Crain, J. Exp. Sci. Environ.

Epidemiol. 2008, 18, 600.[12] L. Claudio, Environ. Health Persp. 2008, 116, A116.[13] B. Bocca, G. Forte, F. Petrucci, S. Costantini, P. Izzo, Sci. Total

Environ. 2009, 407, 2183.[14] X. Li, W. Berger, C. Musante, M. I. Mattina, Chemosphere 2010,

80, 279.[15] R. Baleki, J. Courter, P. Ruiz, B. Sanders, T. Steele, I. Vulic,

‘‘Fielding the Challenges of Artificial Turf Stabilization’’,in Society of Plastics Engineers – International PolyolefinsConference – FLEXPACKCON, 2008, Vol. 1, Houston, TX 2008,p. 596 ff.

[16] A. Grigoriu, V. Blascu, ‘‘Polyolefin Fibers’’, in Handbook ofPolyolefins (Ed. C. Vasile), Marcel Dekker, New York 2000,p. 773 ff.

[17] P. M. Bever, U. Breiner, G. Conzelmann, B. S. von Bernstoff,Chem. Fibers Int. 2000, 50, 176.

[18] M. Haitchi, ‘‘Improvements in Artificial Turf’’, in 44th Inter-national Man-Made-Fibres Congress, Dornbirn 2005.

[19] H. J. Kolitzus, Artificial Turf Surfaces for Soccer, IST Consulting,Eschenz 2007, p 26.

[20] S. Derler, U. Schrade, L. C. Gerhardt, Wear 2007, 263, 1112.




[21] Z. Chen, T. Li, X. Liu, R. Lu, J. Polym. Sci., Part B: Polym. Phys.2005, 43, 2514.

[22] B.-B. Jia, T.-S. Li, X.-J. Liu, P.-H. Cong, Wear 2007, 262,1353.

[23] J. E. Mark, Polymer Data Handbook, 2nd edition Oxford Uni-versity Press, New York 2009.

[24] S. S. Kathiervelu, Synth. Fibres 2002, 31, 11.[25] S. Houis, F. Schreiber, T. Gries, Fibre Table According to P.-A.

Koch: Bicomponent Fibres, Shaker, Aachen 2008.[26] H. J. Koslowski, Tech. Text. 2009, 52, E202.[27] F. Fourne, Synthetic Fibers, Hanser Publishers, Munich

1999.[28] H. H. Yang, ‘‘Polyamide Fibers’’, in Handbook of Fiber Chem-

istry, 3rd edition (Ed. M. Lewin), CRC Press, Boca Raton 2007,p. 31 ff.

[29] M. Jaffe, A. J. East, ‘‘Polyester Fibers’’, in Handbook of FiberChemistry, 3rd edition (Ed., M. Lewin), CRC Press, Boca Raton2007, p. 1 ff.

[30] K. J. Smith, Jr., Polym. Eng. Sci. 1990, 30, 437.[31] J. E. Spruiell, ‘‘Structure Formation During Melt Spinning’’, in

Structure Formation in Polymeric Fibers (Ed. D. R. Salem),Hanser, Munich 2001, p. 5 ff.

[32] A. Ziabicki, Fundamentals of Fibre Formation, John Wiley &Sons, London 1976.

[33] R. Hufenus, F. A. Reifler, K. Maniura-Weber, A. Spierings,M. Zinn, Macromol. Mater. Eng. 2012, 297, 75.

[34] Abaqus/Standard V. 6.8 – User’s Manual, Providence, RhodeIsland, USA

[35] A. Schifferle, A. Neels, O. Papes, A. Dommann, E. Mazza,‘‘Combined testing for MEMS characterization’’, in 24th Euro-sensors Conference, Vol. 5, Linz 2010, p. 878 ff.

[36] F. Ide, A. Hasegawa, J. Appl. Polym. Sci. 1974, 18, 963.[37] R. Hufenus, KunststoffXtra 2011, 4, 27.[38] A. S. McNitt, D. M. Petrunak, T. J. Serensits, ‘‘Temperature

Amelioration of Synthetic Turf Surfaces Through Irrigation’’,in: Proceedings of the IInd International Conference on Turf-grass Science and Management for Sports Fields, Beijing,China, June 24–29, 2007 (¼Acta Horticulturae, Vol. 783,)International Society for Horticultural Science ISHS, Leuven,Belgium 2008, p. 573 ff.

[39] T. J. Serensits, A. S. McNitt, D. M. Petrunak, Proc. Inst. Mech.Eng., Part P: J. Sports Eng. Technol. 2011, 225, 139.

013, 298, 653–663


Design and characterization of a bicomponent melt-spun fiber optimized for artificial turf applications

Documents