Smart Materials and Structures

ACCEPTED MANUSCRIPT

Theory of the tensile actuation of fiber reinforced coiled muscles

To cite this article before publication: Caterina Lamuta et al 2018 Smart Mater. Struct. in press https://doi.org/10.1088/1361-665X/aab52b

Manuscript version: Accepted Manuscript

Accepted Manuscript is “the version of the article accepted for publication including all changes made as a result of the peer review process,
and which may also include the addition to the article by IOP Publishing of a header, an article ID, a cover sheet and/or an ‘Accepted
Manuscript’ watermark, but excluding any other editing, typesetting or other changes made by IOP Publishing and/or its licensors”

This Accepted Manuscript is © 2018 IOP Publishing Ltd.

During the embargo period (the 12 month period from the publication of the Version of Record of this article), the Accepted Manuscript is fully
protected by copyright and cannot be reused or reposted elsewhere.
As the Version of Record of this article is going to be / has been published on a subscription basis, this Accepted Manuscript is available for reuse
under a CC BY-NC-ND 3.0 licence after the 12 month embargo period.

After the embargo period, everyone is permitted to use copy and redistribute this article for non-commercial purposes only, provided that they
adhere to all the terms of the licence https://creativecommons.org/licences/by-nc-nd/3.0

Although reasonable endeavours have been taken to obtain all necessary permissions from third parties to include their copyrighted content
within this article, their full citation and copyright line may not be present in this Accepted Manuscript version. Before using any content from this
article, please refer to the Version of Record on IOPscience once published for full citation and copyright details, as permissions will likely be
required. All third party content is fully copyright protected, unless specifically stated otherwise in the figure caption in the Version of Record.

View the article online for updates and enhancements.

This content was downloaded from IP address 128.111.121.42 on 13/03/2018 at 10:48

Page 1 of 23 AUTHOR SUBMITTED MANUSCRIPT - SMS-106167.R1

1
2
3
4
THEORY OF THE TENSILE ACTUATION OF FIBER REINFORCED COILED MUSCLES
5
6
7
8 C. Lamuta1*, S. Messelot2, S. Tawfick1,3*

pt
9
10
11
12 1
Beckman Institute, University of Illinois at Urbana-Champaign, IL, United States
13

cri
14 2
École supérieure de physique et de chimie industrielles de la ville de Paris, France
15
16 3
Department of Mechanical Science and Engineering, Beckman Institute, University of Illinois - Urbana Champaign, IL,
17
18 United States
19
20 * Corresponding author: clamuta@illinois.edu and tawfick@illinois.edu

us
21
22
23
24 ABSTRACT
25
26 There is a strong need for compact artificial muscles capable of applying large contractile strokes
27
28
29
an
and lift heavy weights. Coiled fibers recently emerged as attractive candidates for these purposes,
owing to their simple construction and the possibility of their thermal, electrical and chemical
30 actuation. An intuitive theoretical understanding of the mechanics of actuation of these muscles is
31
32 essential for the enhancement of their performance and can pave the way for the development of
dM
33 new applications and technologies. In this paper, a complete theoretical model for the tensile
34
35
actuation of fiber reinforced artificial muscles is presented and experimentally validated. The
36 model demonstrates that all muscles made from the same material have a universal behavior,
37 which can be described by a single master curve. It enables the systematic design and
38
39 understanding of coiled muscles for specific performance owing to a comprehensive mathematical
40 correlation among the geometry, materials properties, and actuation. Carbon fibers (CF)/
41
42 polydimethylsiloxane (PDMS) coiled muscles are demonstrated as simple to fabricate yet powerful
43 muscles owing to the availability of high strength CF. In addition to showing excellent agreement
pte

44 with the theoretical models, they can be actuated by joule heating or chemical swelling, lift up to
45
46 12,600 times their own weight, support up to 60 MPa of mechanical stress, provide tensile strokes
47 higher than 25%, and a specific work up to 758 J/kg, the latter is more than 18 times higher than
48
49 that of natural muscles.
50
51
ce

52
53 Keywords: coiled artificial muscles, tensile actuation, theoretical model, carbon fibers,
54
polydimethylsiloxane, twisted yarns
55
56
Ac

57
58
59
60
 AUTHOR SUBMITTED MANUSCRIPT - SMS-106167.R1 Page 2 of 23

1
2
3 INTRODUCTION
4
5
6
Twisted and coiled fibers have been recently proposed as a new class of artificial muscles [1].
7 These lightweight muscles can substitute the traditional and heavy electromagnetic actuators in
8 applications where low weight, fine motion and high contractile work are required.

pt
9
10 A torsional actuation from highly twisted fibers was demonstrated for the first time in 2011 [2].
11 Foroughi et al. showed that a twisted carbon nanotube (CNT) yarn is able to provide a rotation of
12
13
250°/mm when charged electrochemically by immersion into an electrolyte. Subsequently, Lima et

cri
14 al discovered that tensile actuation can also be obtained from CNT yarns [3]. In this case, the yarn
15 was overtwisted until a spontaneous formation of coils was observed, and when the ends are
16
17 constrained from rotation but allowed to axially move, a tensile contraction around 7% was
18 measured when the material was heated at incandescent temperatures. The addition of a volume
19
20
expanding guest, such as paraffin wax [3] or silicon rubber [4], was later adopted to allow the

us
21 tensile actuation at lower temperatures. In particular, coiled CNTs yarns infiltrated by silicone
22 rubber, showed a tensile actuation of 34% when electro-thermally actuated, and 50% when
23
24 actuated by swelling [4]. Chemical swelling generally allows larger actuation strains because some
25 polymers can swell up to 300%, albeit on the cost of actuation speed. Although such performances
26
27
28
29
an
are promising, the application of twisted and coiled CNT artificial muscles is limited by the high
cost of fabrication involving aligned CNT yarns. In order to overcome this practical limit, polymeric
fibers were proposed as twisted and coiled artificial muscles in 2014 [5]. These muscles can be
30
31 easily realized by highly twisting low cost and common materials like sewing thread or fishing
32 lines, and can provide tensile actuation higher than 30% when heated to 50°C.
dM
33
34 Recent works on the development of new materials [6, 7] and applications [8-10] for artificial
35 muscle fibers demonstrate the growing interest of the scientific community in this topic. The
36
theoretical understanding of the actuation mechanism of these muscles is then fundamental for
37
38 the improvement of their performances as well as for the design of attractive and sophisticated
39 technologies.
40
41 The first theoretical model for the torsional and tensile actuation of twisted and coiled muscles
42 respectively, was proposed by Haines et al. [5]. In particular, the untwisting (and then the torsional
43
pte

stroke) is considered the consequence of the anisotropic volumetric expansion that the nylon fiber
44
45 experiences during the heating process. The authors proposed a geometric relation to relate the
46 spontaneous untwisting to the change of the geometrical parameters of the fiber (i.e., the bias
47
48 angle, the diameter, and the length). The spontaneous untwisting was also considered the cause
49 of the tensile actuation of coiled fibers [5]. A simple geometric equation that relates the tensile
50
stroke to the untwisting, the number of coils, and the length of the fiber, is derived based on
51
ce

52 Love’s helix equation in his mathematical treatise on elasticity [11]. A single helix equivalent model
53 was later developed by Aziz et al [12] to relate the kinematics of the untwist to the volume change
54
55 caused by heating and demonstrate that the torsional stroke depends only on the inserted twist
56 and is independent on fiber diameter.
Ac

57
The existing theoretical models described above treat the geometry and kinematics of the
58
59 actuation with no correlation between the performance, the loading conditions, and the
60 mechanical properties of the material. In the present work, we focus on the tensile actuation of
Page 3 of 23 AUTHOR SUBMITTED MANUSCRIPT - SMS-106167.R1

1
2
3 coiled muscles proposing a comprehensive and intuitive theoretical model to explain the physical
4
5 mechanism of their actuation and the limits of their performance. The model describes first the
6 mechanics of coiling as related to the geometry and material properties of the fiber. The model
7
8
goes beyond simple kinematics by relating the performance of the muscles not only to the

pt
9 geometric parameters of the coil, but also to the loading conditions and the thermo-electro-
10 mechanical response of the materials. The starting point for the development of the model is the
11
12 equilibrium of work and energy during the coiling of elastic rods, as proposed by Love [11], and
13 revised by Thompson et al. [13]. We demonstrate that the muscle contraction (i.e., the tensile

cri
14
15
actuation) is caused by the increase in the radius of the yarn, due to thermal expansion or solvent
16 absorption. The increase of the radius causes an increment of the bending and the torsional
17 stiffness. The “new” muscle’s equilibrium extension is offset by an increment of the coil angle, and
18
19 the transition from the previous equilibrium position to the new equilibrium configuration of the
20 stiffer coil leads to the contraction of the muscle.

us
21
22
The proposed model is validated by experimental tests performed on novel carbon fibers (CF)/
23 Polydimethylsiloxane (PDMS) coiled muscles. The effect of the geometry of the twisted yarn is also
24 considered, so that our methodology can be applied for the design and the optimization of any
25
26 fiber reinforced coiled muscle. These muscles can be actuated by joule heating or solvent swelling.
27
28
29
an
They support up to 60 MPa of mechanical stress and provide an actuation of 25% and a specific
work of 758 J/kg when an input voltage of 1.64 V/cm is applied.
30
31
32
dM
33 NOMENCLATURE
34
35
36 As = cross sectional area of the fiber/composite yarn
37 B = bending stiffness of the fiber/composite yarn
38 Bf = bending stiffness of the coiled muscle (i.e. coiled fiber/composite yarn)
39 after the actuation
40
Bi = bending stiffness of the coiled muscle before the actuation
41
42 β = coil angle of the coiled muscle
43 βf = coil angle of the coiled muscle after the actuation
pte

44 βi = coil angle of the coiled muscle before the actuation

45 C = torsional stiffness of the coiled muscle
46
47
Cf = torsional stiffness of the coiled muscle after the actuation
48 Ci = torsional stiffness of the coiled muscle before the actuation
49 cp = specific heat of the coiled muscle
50 cpfiber = specific heat of the fiber of the composite yarn
51
ce

cpmatrix = specific heat of the matrix of the composite yarn

52
53 D = total axial end shortening of the fiber/composite yarn during coiling
54 l = axial displacement obtained during the actuation
55 Ex = Young’s modulus of the composite yarn along the axial direction
56 Gyz = shear modulus of the composite yarn in the transverse section
Ac

57
58 h = convection coefficient between the coiled muscle and the surrounding air
59 Ic = electric current
60 L = pitch of the coiled muscle
M = twisting moment applied to the upper extremity of the straight
 AUTHOR SUBMITTED MANUSCRIPT - SMS-106167.R1 Page 4 of 23

1
2
3 fiber/composite yarn
4
5
m = mass of the coiled muscle
6 Mc = critical moment at the beginning of the spontaneous coiling
7 mfiber = mass of the fibers contained into the composite yarn
8 mmatrix = mass of the matrix of the composite yarn

pt
9 N = number of coils of the coiled muscle
10
11 n = total number of turns applied during coiling
12  = efficiency (ratio between the output mechanical energy and the input
13 electrical energy)

cri
14 r = total end rotation of the fiber/composite yarn in radians per unit length
15
R = electrical resistance of the coiled muscle
16
17 rf = radius of the fiber/composite yarn after the actuation
18 ri = radius of the fiber/composite yarn before the actuation
19 ρ = coil radius of the coiled muscle
20 S = initial length of the straight fiber/composite yarn

us
21
22 T = tensile load applied to the lower extremity of the fiber/composite yarn
23 T(t) = temperature of the coiled muscles at the time t
24 Tamb = ambient temperature
25 Tc = critical tensile load at the beginning of the spontaneous coiling
26
27
28
29
τi
V
V(t)
= internal twist during coiling
an
= total potential energy of the fiber/composite yarn
= voltage applied for a time t
30 VB = strain energy of bending
31 VC = strain energy of torsion
32
dM
33 VM = potential energy of the moment M
34 VT = potential energy of the tensile load T
35
36
37
38 METHODS
39
40 Theoretical model
41
42 When a straight and elastic fiber is highly twisted, a spontaneous formation of coils can be
43
pte

44 obtained if a critical condition of coiling instability is reached. The understanding of this

45 phenomenon is crucial for the development of a mechanics model, able to explain the actuation
46
47
mechanism of coiled artificial muscles. We use the nonlinear formulation of Love [11] to describe
48 the coiling procedure of a highly twisted fiber. The starting configuration is represented by a
49 straight fiber with length S and radius ri, loaded at its ends by a twisting moment M, and a tensile
50
51 load T, as shown in Figure 1. The fiber is assumed to be linearly elastic exhibiting no axial or shear
ce

52 deformability.
53
54
Under combined tension and moment, the straight fiber becomes unstable and form a
55 spontaneous coil when the external loads M and T reach a critical value, Mc and Tc respectively,
56 such that:
Ac

57
58
59 𝑀𝑐2 = 4𝐵𝑇𝑐 (1)
60
Page 5 of 23 AUTHOR SUBMITTED MANUSCRIPT - SMS-106167.R1

1
2
3 where B is the bending stiffness of the fiber B=EI, where E is the Young modulus of the material
4
5 and I the second moment of area. The critical condition expressed in Equation 1 can be easily
6 obtained by the linear eigenvalue analysis of Love [11] or Timoshenko [14].
7
8
First, the helical geometry of the spontaneously formed coil can be described by:

pt
9
10 2𝜋𝜌
11
𝐿= (2)
tan 𝛽
12
13
𝑆 sin 𝛽

cri
14 𝜌= (3)
15 2𝜋𝑁
16
17 where L is the coil pitch, ρ its radius, β is the coil angle, and N the number of coils formed
18 spontaneously, which will be described next. The spontaneous contraction of the fiber, indicated
19
20 as D in Figure 1, is a geometric function of the coil angle and can be calculated as follows:

us
21
22 𝐷 = 𝑆(1 − cos 𝛽) (4)
23
24
25
26
27
28
29
an
30
31
32
dM
33
34
35
36
37
38
39
40
41
42
43
pte

44
45 Figure 1: Schematic representation of the fiber geometry and the spontaneous coiling due to overtwisting.
46
47
48
49 To determine the geometry of the spontaneously formed coil, β and N must be determined from
50
51
the mechanics of coil formation. The total energy of the system can be written as:
ce

52
53 𝑉 = 𝑉𝑇 + 𝑉𝑀 + 𝑉𝐵 + 𝑉𝐶 (5)
54
55 where VT, VM, VB and VC are the potential energy due to the tensile load T, the external moment
56
M, the bending, and torsion respectively. All the energy terms listed in Equation 5 can be written
Ac

57
58 in relation to the external loads, the mechanical properties of the material, and the geometric
59
parameters of the coils as follows:
60
 AUTHOR SUBMITTED MANUSCRIPT - SMS-106167.R1 Page 6 of 23

1
2
3
𝑉𝑇 = 𝑇 𝑆 (1 − cos 𝛽) (6)
4
5
6 𝑉𝑀 = −𝑀 𝑆 (𝜏𝑖 + sin 𝛽/𝜌) (7)
7
8 1
𝑉𝐵 = 𝐵 𝑆 𝑠𝑖𝑛4 𝛽/𝜌2

pt
9 (8)
2
10
11
1
12 𝑉𝐶 = 𝐶 𝑆 [𝜏𝑖 + sin 𝛽 cos 𝛽/𝜌]2 (9)
13 2

cri
14
15 where τi the internal twist, and C is the torsional stiffness, C=GJ, where G is the shear modulus of
16 the material and J the polar second moment of area for a circular fiber. When the derivatives of
17
18 the total energy with respect to the coil angle, the coil radius, and the internal twist, are set equal
19 to zero, the equilibrium condition in the post-buckled state (i.e., after the formation of coils) can
20

us
21 then be found. Under the minimum energy configuration, the following equations relating the
22 fiber’s material and geometry to the coiling, and as can be shown later the mechanics of its
23
actuation, can be obtained:
24
25
26 1 𝑟 𝐵𝐶
𝛽 = cos−1 [
27
28
29
an
(
𝐵 − 𝐶 √𝐵𝑇
− 2𝐶) − 1] (10)

30 𝑆 𝑇
31 𝑁= √ (11)
32
2𝜋 𝐵
dM
33
34 The parameter r is the twist per unit length defined as total end rotation of the fiber in radians per
35
unit length, calculated as follows:
36
37
38
2𝜋n
𝑟= (12)
39 𝑆
40
41 were n is the total number of turns applied during the twisting procedure. Equation (10) is very
42 useful because it allows the design of the final geometry of the coil, starting from a straight fiber
43
pte

44 (with known dimensions and mechanical properties), by choosing the value of the applied tensile
45 load T and number of turns n. We note from Equation 11 that the number of coils N is a function
46
47
of the tensile load T and the bending stiffness B only.
48 This simple equation can also be used to explain why a spontaneously coiled artificial muscle is
49 able to provide a tensile actuation when it receives an input such as simple heating, Joule heating,
50
51 or solvent absorption. The effect of all the different input typologies mentioned above is the
ce

52 volumetric expansion of the fiber. Such expansion can be anisotropic for some materials such as
53
54 carbon nanotubes or carbon fibers, which experience a large expansion along the radius of the
55 fiber and a small contraction along the longitudinal axis. This phenomenon is related to the
56 anisotropy of the graphite crystallites, which exhibit different properties along the directions
Ac

57
58 parallel and perpendicular to the basal plane [15]. Also some oriented crystalline polymers such as
59 nylon show a similar anisotropy during volumetric expansion [16].
60
Page 7 of 23 AUTHOR SUBMITTED MANUSCRIPT - SMS-106167.R1

1
2
3 This anisotropic volumetric expansion is the key factor for the development of coiled artificial
4
5 muscles. In the model developed by Haines et al. [5] it was in fact considered the cause of the
6 spontaneous untwisting, which was in turn proposed as the cause of the tensile actuation.
7
8
The relation between the volumetric expansion, the untwisting, and the resulting actuation, can

pt
9 be explained in light of Equation (10). The effect of the anisotropic volumetric expansion can be
10 easily considered in Equation (10) as an increment of the radius of the fiber from ri to rf, while the
11
12 length of the fiber S can be kept constant for the sake of simplicity (compared to the radius
13 increase, the shortening of the fiber length can in fact be considered negligible for all the materials

cri
14
15
mentioned above, mostly for fiber reinforced coiled muscles like carbon fiber-based or CNT-based
16 muscles). The radius increment causes then a significant increase of the bending and torsional
17 stiffness B and C, as well as a variation of the coil angle β, from an initial configuration i to a final
18
19 configuration f, as described by the following equations:
20

us
𝜋
21 𝐵𝑖,𝑓 = 𝐸 4 𝑟𝑖,𝑓 4 (13)
22
23 𝜋
24 𝐶𝑖,𝑓 = 𝐺 2 𝑟𝑖.𝑓 4 (14)
25
26
27
28
29
𝛽𝑖,𝑓 = cos−1 [
1
(
an
𝑟 𝐵𝑖,𝑓 𝐶𝑖,𝑓
𝐵𝑖,𝑓 − 𝐶𝑖,𝑓 √𝐵𝑖,𝑓 𝑇
− 2𝐶𝑖,𝑓 ) − 1] (15)
30
31
32 As a consequence, also the coil parameters N, L, and ρ will be subject to a variation (according to
dM
33 Equations (11), (2), and (3), respectively), as shown in Figure 2.
34
35
36
37
38
39
40
41
42
43
pte

44
45
46
47
48
49
50
51
ce

52
53
54
55
56 Figure 2: (a) Calculated change in the coil angle β, coil radius ρ, number of turns N, pitch L, and displacement l, for
Ac

57 increasing values of the radius of the fiber (r/r = (rf - ri) / ri.), obtained using the equations in the main text; and (b)
58 schematic representation of the tensile actuation mechanism. The global effect of the radius increase on the coil
59
parameters in shown. The top end of the coil is fixed, while the bottom end can only translate along the vertical
60
direction (in order to avoid the untwisting), and a tensile load T is applied on it.
 AUTHOR SUBMITTED MANUSCRIPT - SMS-106167.R1 Page 8 of 23

1
2
3
4
5
The result of the anisotropic volume expansion is then a “new” stiffer coil, characterized by a
6
7 different equilibrium position. The direct consequence of this new equilibrium condition is the
8 shortening of the coil by a displacement l, leading to tensile actuation. The tensile stroke, that

pt
9
10 corresponds to the displacement of the bottom end, can be calculated by simple geometrical
11 considerations as follows:
12
13
∆l = 𝑆(cos 𝛽𝑖 − cos 𝛽𝑓 ) (16)

cri
14
15
16 Equations (13-16) can then be used to predict the tensile actuation of any coiled fiber that
17 experiences an anisotropic volume expansion. In particular, the initial geometry of the coil can be
18
19 controlled using Equation (15), by knowing the mechanical properties of the material (i.e., E, G),
20 the initial dimensions (i.e., ri, S), and controlling the external loads (i.e., T and n). A suitable input

us
21
stimulus (thermal, electrical, or by swelling) can then be chosen in order to provide a radius
22
23 increment able to produce a desired output displacement.
24
25
26 Anisotropy of the fiber reinforced muscles
27
28
29
an
The model proposed above is very useful to design the tensile actuation of coiled artificial
muscles. However, it refers to a material with isotropic mechanical properties. In Equations (13-
30
31 14) in fact only a single value for the Young’s modulus E and the shear modulus G is considered.
32 In the work by Lima et al. [3, 4], it is shown how coiled CNT yarns are able to provide large tensile
dM
33
34
strokes at lower temperatures if infiltrated with wax or silicon rubber, due to the higher
35 volumetric thermal expansion coefficients of the matrix. More generally, a coiled artificial muscle
36 can be produced with any stiff fiber able to confine a guest material as a matrix to anisotropic
37
38 volume expansion, and such consideration pushes the research toward the development of new
39 materials combinations [7, 17]. In this scenario, a complete theoretical model that takes into
40
41
account the mechanical anisotropy of a fiber reinforced coiled muscle is highly needed.
42 We consider herein the generic case of a fiber reinforced composite whose initial configuration
43
pte

before the coiling procedure (i.e., the straight configuration on the left of Figure 1) is shown in
44
45 Figure 3 (c). For the sake of clarity, since the discussion will be focused on an anisotropic
46 composite material in the next sections of the paper, the term “fiber” used during the description
47
48
of the theoretical model will be replaced by the term “composite yarn”.
49 In particular, the composite material is characterized by long fibers embedded into a matrix and
50 “frozen” in a twisted configuration (the production process and the motivations behind it are
51
ce

52 described in the next session). In Figure 3 (b) a schematic representation of the initial composite
53 straight yarn is shown. Before twisting insertion, the on-axis reference system 1,2,3 coincides with
54
55 the global system x, y, z, as shown in Figure 3 (a). After the twisting procedure, an angle α is
56 generated between the fibers (axis 1) and the x axis. Such angle coincides with the external twist
Ac

57 angle of the fibers yarn and is defined positive when is measured in a counter-clock-wise direction
58
59 from the axis 1 to the axis x (for such reason the twist angle in Figure 3 (c) has been indicated as –
60 α). If during the coiling procedure the tensile load T is applied along the x direction and the
external moment M around the x axis, Equations (13-14) have to be modified as follows:
Page 9 of 23 AUTHOR SUBMITTED MANUSCRIPT - SMS-106167.R1

1
2
3 𝜋
𝐵𝑖,𝑓 = 𝐸𝑥 4 𝑟𝑖,𝑓 4 (17)
4
5
𝜋
6 𝐶𝑖,𝑓 = 𝐺𝑦𝑧 2 𝑟𝑖,𝑓 4 (18)
7
8
Ex in this case is the relevant modulus as the resistance to bending B of the muscle is the

pt
9
10 resistance to extension in the x-direction above and below the neutral axis of the coiled muscles.
11 The Naik et al. model [18] is used to calculate Ex and Gyz from the known mechanical properties of
12
13 the matrix and the fibers, as described in the Appendix section.

cri
14
15
16
17
18
19
20

us
21
22
23
24
25
26
27
28
29
an
30
31
32
dM
33 Figure 3: Schematic representation of the straight yarn (a) before and (b) after twisting; (c) optical image of CF/PDMS
34 straight yarn with frozen twisted carbon fibers after the curing of the PDMS.
35
36
37
38 Muscle fabrication and experimental setup
39
40 The theoretical model proposed in the previous section was experimentally validated for CF/PDMS
41
coiled muscles. The fabrication process adopted to produce one coil is extremely quick and
42
43 straightforward and can be reproduced in any environment with no needs for sophisticated
pte

44 equipment or stringent experimental requirements.

45
46 Commercial Thornel-T300 PAN-based carbon fibers (CF), supplied with 1% UC.309 epoxy-
47 compatible sizing, were used. One CF tow (containing 4500 fibers) is first dipped into uncured
48
49
Sylgard 184 PDMS, diluted with hexane (2:1 by weight). The presence of the hexane is essential to
50 decrease the viscosity of the PDMS and obtain a more homogeneous distribution of the PDMS
51
ce

inside the CF tow. At the end of the dipping stage, the tow is twisted by using a simple drill. The
52
53 composite yarn with twisted carbon fibers (with both ends tethered) is then placed in the oven at
54 50°C for 24 h in order to cure the PDMS.
55
56 The twisting procedure before the curing of the PDMS is necessary to eliminate the excessive
Ac

57 amount of liquid PDMS and obtain an initial straight composite yarn with a homogeneous shape
58 and a constant radius as shown in Figure 3 (c). In particular, 240 turns/m were applied during the
59
60 pre-twisting procedure, which led to a composite yarn with a PDMS content of 4̴ 0%wt and radius
of 2̴ 00μm. During twisting, the PDMS redistributed filling the gaps between the fibers and the
 AUTHOR SUBMITTED MANUSCRIPT - SMS-106167.R1 Page 10 of 23

1
2
3 excess deposits on the outside of the composite yarn. One can observe that the curing of the
4
5 PDMS froze the twisting of the carbon fibers at a certain angle α. The measurement of such angle
6 is important for the calculation of the mechanical properties of the composite, as explained in the
7
8
previous section and in the Appendix of the present work.

pt
9 When the PDMS is cured, the straight composite yarn was then highly twisted until it was fully
10 coiled. Figure 4 (a) shows an example of fully coiled CF/PDMS muscle.
11
12 The experimental setup used for the electrical actuation of the coiled muscle is shown in Figure 4
13 (b). A DC input voltage was applied at the coil ends in order to induce the heating of the composite

cri
14
15
yarn and the radius increase, causing of the tensile actuation. The tensile stroke was captured by a
16 movie camera and the displacement was measured by the analysis of movie frames. The top end
17 of the coil was fixed, while a weight was attached to the bottom end in order to apply a targeted
18
19 tensile load. A horizontal rigid rod was attached to the weight and place in contact with two
20 vertical rails in order to avoid its rotation. A ruler with a 0.5 mm resolution was attached to one of

us
21
22
the vertical surfaces to facilitate the displacement measurements. An optical photograph of the
23 experimental setup can also be found in Figure S1.
24
25
26
27
28
29
an
30
31
32
dM
33
34
35
36
37
38
39
40
41
42
43
pte

44
45
46
47 Figure 4: (a) coiled CF/PDMS composite yarn; (b) schematic representation of the experimental setup used for the
48 electrical actuation. The composite yarn is constrained from rotation at the top fixture point, as well as at the bottom
49 using a horizontal bar constrained between frictionless rails. A weight is applied at the bottom. A voltage is applied
50 between the ends of the muscle.
51
ce

52
53
54
55 RESULTS AND DISCUSSION
56
Ac

57 Experimental validation for electro-thermally actuated muscles

58
59 The ability of the proposed model to predict the tensile actuation of fiber reinforced coiled
60
muscles is tested on the produced CF/PDMS composite yarn. A straight yarn with a radius ri =192
Page 11 of 23 AUTHOR SUBMITTED MANUSCRIPT - SMS-106167.R1

1
2
3 μm and an initial length S = 16.5 cm is highly twisted in order to induce the spontaneous coiling.
4
5 One hundred and fifty turns n = 150 and a tensile load T = 0.44 N are applied. Under such loading
6 conditions, a contraction of the coil equal to 9.8 cm and a total number of coils N = 41 are
7
8
measured (see Figure S2 (b)) and they are in perfect agreement with the theoretical predictions of

pt
9 D = 9.7 cm and N = 39.24 obtained from Equation (4) and (11) respectively (see Figure S2 for
10 calculation details), confirming the capability of the theoretical model to evaluate the final
11
12 configuration of the composite yarn at the end of the coiling procedure.
13 Next, we demonstrate the ability of the proposed model to predict the tensile actuation of this

cri
14
15
coiled muscle. A DC input voltage is applied to the extremities of the coil to induce an increment of
16 the radius of the yarn by Joule heating and relate this increment to the output coil contraction.
17 This procedure allows to experimentally obtain the plot l-(r/r) of Figure 2 (a) and then a direct
18
19 comparison with the theoretical predictions.
20 A in situ measurement of the radius increment during the tensile actuation process is extremely

us
21
22 challenging, due to the untwisting and the translation experienced by the muscle, combined to the
23 small entity of the increment itself. On the other hand, using an infrared camera, the temperature
24 of the coil can be easily measured during the actuation. We combine this in situ temperature
25
26 measurement with an ex situ measurement of the radius increment as function of heating for a
27
28
29
an
twisted fiber without coiling. In the latter experiment, a straight yarn is placed under the optical
microscope, and its expansion under Joule heating is recorded with a movie camera in order to
30 measure the radius increment from the analysis of movie frames. The temperature is measured by
31
means of a thermocouple in contact with the yarn. The correlation between the temperature and
32
dM
33 the radius increment is then obtained and it is shown in Figure 5 (a). The slope of the initial linear
34 region in the plot is equal to 3x10-4 °C-1, that corresponds to the linear thermal expansion
35
36 coefficient CTE of the Sylgard 184 [19].
37 Next, the temperature is measured during the actuation of the coiled muscle and the
38
experimental r/r value is calculated from the plot of Figure 5 (a). In particular, the temperature
39
40 during the tensile actuation of the coil is recorded by a thermal camera (inset in Figure 5 (b)).
41
42
43
pte

44
45
46
47
48
49
50
51
ce

52
53
54
55
56
Ac

57
58 Figure 5: (a) Temperature versus radius increment for a straight CF/PDMS yarn under Joule heating. The inset shows a
59 frame of the movie recorded during the expansion under the microscope; (b) Comparison between experimental (red
60 points) and theoretical results for the tensile actuation of a coiled CF/PDMS muscle when a DC voltage of 7.5 V is
 AUTHOR SUBMITTED MANUSCRIPT - SMS-106167.R1 Page 12 of 23

1
2
3 applied for 9 seconds. The inset shows a frame related to the temperature measurement by a thermal camera during
4 the tensile actuation (a temperature of 130°C was measured after 9 seconds).
5
6
7
The r/r experimental values are then correlated to the values of the displacement l measured
8 during the electrical actuation (by means of the analysis of movie frames, as explained in the

pt
9 previous section and Figure 2 (a)), as shown in Figure 5 (b) (red points). A perfect agreement
10
11 between the experimental results and the theoretical prediction can be observed. The theoretical
12 plot was obtained from Equations (15-18) by increasing the radius of the yarn. The geometrical
13
parameters and loading conditions listed above (i.e., ri =192 μm; S = 16.5; n = 150; T = 0.44 N) are

cri
14
15 used in the calculations. The calculation of the mechanical properties Ex and Gyz is described in the
16 Appendix (the value of the twist angle α and the CF volumetric fraction vf, needed for the
17
18 mechanical properties evaluation, were measured by optical images and SEM analysis
19 respectively, as shown in Figure S2).
20
The experimental data of Figure 5 (b) corresponds to electrical actuation performed using a DC

us
21
22 input voltage of 7.5 V, for 9 seconds. A temperature of 130°C is measured by the thermal camera
23 after 9 seconds (inset in Figure 5(b)).
24
25 The theoretical prediction of the tensile displacement can be further generalized if we consider
26 the simple equation that rules the Joule heating phenomenon and that is reported below [20]:
27
28
29
an𝐶 𝑝𝑡 𝐶
𝑇(𝑡) = 𝑒 −𝑝𝑡 ( 𝑒 + (𝑇𝑎𝑚𝑏 − )) (19)
30 𝑝 𝑝
31
32
dM
33
were:
34
35 ℎ 𝐴𝑠
𝑝= (20)
36 𝑚 𝑐𝑝
37
38 𝑉(𝑡)2 1
39 𝐶=( + ℎ 𝐴𝑠 𝑇𝑎𝑚𝑏 ) ( ) (21)
40 𝑅 𝑚 𝑐𝑝
41
42 T is the instantaneous temperature at time t when a voltage V is applied on a yarn of mass m,
43
pte

44
electrical resistance R, section As, and specific heat cp. h is the convection coefficient between the
45 yarn and the surrounding air at temperature Tamb that in conditions of free convection is equal to
46 1W/m2°C. The product m cp in Equation (20) corresponds to the properties of the composite
47
48 material where m cp = mmatrix cpmatrix + mfiber cpfiber.
49 Equation (19) can be used to predict the temperature reached by the coil if an input voltage V is
50
51
applied for a time t, with no need for temperature measurements by a thermal camera during the
ce

52 coil actuation. In particular, according to Equation (19), and using the geometrical and thermal
53 properties of the tested coil (measured resistance R = 40 Ω, As = π (192 μm)2, mPDMS = 0.041 g,
54
55 cpPDMS = 1 J/g°C, mCF = 0.046 g, cpCF = 1.46 J/g°C), a value of V equal to 7.1 V has to be used for 9
56 seconds to reach a temperature T(9) = 130°C. Such value is extremely close to the used value of
Ac

57
58
7.5 V (the used voltage is slightly higher because Equation (19) does not consider the heat loss due
59 to the electrical wires used for the actuation), confirming the suitability of Equation (19).
60
Page 13 of 23 AUTHOR SUBMITTED MANUSCRIPT - SMS-106167.R1

1
2
3 Therefore, using Equations (15-21), the tensile performances of an electrically activated fiber-
4
5 reinforced coiled muscle can be directly predicted from the geometrical, mechanical, thermal, and
6 electrical properties of the composite material, according to the applied loading conditions. Figure
7
8
6 shows how the comprehensive model proposed in the present work led to the evaluation of the

pt
9 tensile stroke of a coiled muscle actuated with an input voltage of 10 V for 8 seconds.
10 The order of the procedure described in Figure 6 can be easily adapted to design the size of a
11
12 coiled muscle, starting from specific values of displacement l and load T required for a particular
13 application.

cri
14
15
16
17
18
19
20

us
21
22
23
24
25
26
27
28
29
an
30
31
32
dM
33
34
35
36
37
38
39
40
41
42
43
pte

44
45
46
47 Figure 6: Complete procedure for the prediction of the tensile actuation by using the model proposed in the present
48 work. Output results refer to a coiled muscle with the following input parameters: V=10V, t=8sec, ri=183μm, S=15cm,
49 Ex=467MPa, Gyz=22MPa, h=1, mcp=0.1082 J/°C, R=40Ω, T=0.52N, r=8168 turns/m.
50
51
ce

52 Tailoring the performance of coiled muscles

53
54 The complete theoretical model proposed in this work allows the tailoring of the muscle size and
55
56 performance. In particular, the tensile actuation of a coiled muscle composed by a yarn of fibers
Ac

57 and an expanding guest material (like the CF/PDMS muscles analyzed in this work) can be limited
58
59
by two factors: the limited expansion of the guest material, or the coil-coil contact. The first limit is
60 strictly related to the properties of the guest material that, above its degradation temperature, it
is not able to provide an additional volume expansion. This is why muscles actuated by swelling
 AUTHOR SUBMITTED MANUSCRIPT - SMS-106167.R1 Page 14 of 23

1
2
3 have higher actuation strains, as they are able to swell significantly more than heat-induced
4
5 expansion without degradation. On the other hand, the coil-coil contact is a geometrical limit:
6 when adjacent coils come into contact, the muscle can no longer contract, even if the guest
7
8
material is below its degradation temperature and is able to cause an additional increment of the

pt
9 radius of the fiber. The interplay between these two limits and their effect on the muscles
10 performance can be easily analyzed using the model proposed in this paper.
11
12 PDMS usually starts to degrade at temperatures higher than 200°C [21]. We observed clear smoke
13 associated with degradation of the used Sylgard 184 for temperatures higher than 230 °C, as

cri
14
15
measured by the infrared camera. Since the expansion of the guest PDMS can no longer be
16 exploited after its evaporation, the temperature of 230°C was considered the limit working
17 temperature of the produced CF/PDMS muscles. From the plot in Figure 5 (a) this temperature
18
19 corresponds to a radius increment r/r equal to 0.1. Using this value of the radius increments, the
20 theoretical plots in Figure 7 are obtained by means of Equations (15-18).

us
21
22 Figure 7 (a) and (b) show how the initial coil angle βi and the final coil angle βf vary for different
23 values of tensile stress and inserted twist. With a constant twist r (and then a constant number of
24 turns, n), the value of the initial coil angle decreases with increasing values of the applied tensile
25
26 stress. When the applied load is larger, the coil is more stretched, and the coil angle is smaller. At
27
28
29
an
high stress and small number of turns, βi reaches the zero value (black curve in Figure 7 (a)),
corresponding to a straight yarn configuration. The increment in the number of turns causes
30 instead an increase in the initial coil angle, and, with a very high number of turns and a small
31
applied stress, the other limit configuration of the coil-coil contact, with a coil angle equal to 90°
32
dM
33 (dashed curves in Figure 7) can be reached. Similar considerations can be deduced for the final
34 angle βf related to the condition of maximum expansion of the PDMS. The effect of the variation
35
36 of the initial and final coil angles on the performance of the muscles can be observed in Figure 7
37 (c) and (d). Figure 7 (c) clearly shows that the maximum tensile actuation of the stress/turns
38
configurations represented by the black and the red curves (i.e., 17 % and 42 % respectively) is
39
40 only limited by the exploitable expansion of the PDMS, whereas the coil-coil contact limits the
41 actuation in the other two cases (blue and magenta curves). It is worth noting that the maximum
42
43 tensile actuation obtained at the coil-coil contact configuration is constant for different
pte

44 stress/turns combinations. In particular, it is equal to 46 % for a CNTs/PDMS yarn with the

45
following properties: ri=183μm, S=15cm, Ex=920MPa, Gyz=23MPa. Figure 7 (c) also demonstrates
46
47 how the model is able to indicate which value of the tensile load has to be used to reach a desired
48 tensile actuation when a certain number of turns r is applied. This guides the choice of the muscle
49
50 diameter and length.
51
ce

The results of Figure 7 can be further generalized and a single “master curve”, describing the
52
universal performance of muscles from the same material can be obtained. Figure 8(a) and (b)
53
54 show how the curves of Figure 7 (c) and (d), related to different number of turns, can collapse in a
55 single curve if the stress/r2 ratio (where r is the twist per unit length) is considered instead of the
56
simple stress along the horizontal axis. More specifically, the trend of the specific work can be
Ac

57
58 generalized considering a reference number of turns r0, as shown in Figure 8(b).
59
60
In Figure 8(a) and (b) the experimental results for two different muscles are also shown and good
agreement with the theoretical prediction can be observed. However, a clarification has to be
Page 15 of 23 AUTHOR SUBMITTED MANUSCRIPT - SMS-106167.R1

1
2
3 made about the maximum value obtained for the experimental tensile actuation. Both the tested
4
5 muscles were coiled using an inserted twist r = 8168 turns/m. According to the theoretical
6 predictions of the red curve in Figure 7 (c), a maximum tensile actuation bigger than 40%
7
8
(obtained in conditions close to the coil-coil contact) is expected when a stress of 2 MPa is applied

pt
9 to a muscle with an inserted twist of 8168 turns/m. However, the maximum tensile actuation
10 measured for these muscles never exceeded a value of 25% (as shown in Figure 8 (a)). This
11
12 discrepancy between the theoretical and the experimental maximum tensile actuation results
13 from the fact that Equation (16) assumes that the coiled muscle is modeled as an elastic line with

cri
14
15
zero thickness. Due to the thickness (i.e. the radius) of the yarn, the real coil-coil contact is
16 reached for coil angles lower than 90 deg, and, as a consequence, the real maximum tensile
17 actuation is lower than the maximum theoretical one.
18
19
20

us
21
22
23
24
25
26
27
28
29
an
30
31
32
dM
33
34
35
36
37
38
39
40
41
42
43
pte

44
45
46
47
48
49 Figure 7: Calculated performance of coiled muscles. (a) initial coil angle, (b) final coil angle, (c) tensile actuation, and
50 (d) specific work, calculated for different values of stress and inserted twist r, in the condition of maximum expansion
51
ce

of the PDMS (related to a limit temperature of 230°C and a r/r = 0.1). The plot refers to a coiled CF/PDMS muscles
52
53 with ri=183μm, S=15cm, Ex=920MPa, Gyz=23MPa.
54
55 As shown in Figure 7 (c) and (d), the fabricated CF/PDMS coiled muscles are able to provide a
56 maximum tensile actuation up to 25 % and a maximum specific work of 0.758 kJ/kg (this value can
Ac

57
58 be calculated from Figure 8 (b) considering that the experimental data refer to an inserted twist r=
59 8168 turns/m). Moreover, they are able to lift more than 12,600 times their own weight and
60
support up to 60 MPa of mechanical stress.
 AUTHOR SUBMITTED MANUSCRIPT - SMS-106167.R1 Page 16 of 23

1
2
3 The value of tensile actuation is higher than the maximum value obtained for electrically actuated
4
5 coiled Fermat CNT yarns infiltrated with wax ( ̴3%) [3], and comparable to the tensile strokes of
6 thermally actuated coiled nylon fibers [5] and electrically actuated CNTS yarns infiltrated by
7
8
silicone rubber ( ̴30%) [4]. However, we would like to point out that the performances of CNT

pt
9 muscles in the works by Lima et al. [3, 4] were optimized using dynamic electrical actuation which
10 could enable higher instantaneous expansion, as well-as dynamic inertial effects in the load. The
11
12 CF/PDMS muscles studied in this work are tested by using a DC electrical input. In the future, we
13 plan to study their dynamic response thoroughly.

cri
14
15
Furthermore, a tensile actuation up to 58% is recorded for the CF/PDMS coils when very low
16 tensile loads and high number of turns are applied. Under such loading conditions, tertiary coiling
17 is encountered, and the model is not valid. The starting configuration of the muscle is very close to
18
19 the coil-coil contact (as shown in Figure 7 (a)) and when the coils get in contact, the fiber
20 continues to increase its radius until the muscle folds back on itself due to the low applied stress

us
21
22
(see Figure S3). Similar results, with tensile strokes up to 34% were also measured when the
23 CF/PDMS muscles were actuated by swelling (see Supplementary movie 1). The PDMS swelling
24 was induced by delivering liquid hexane on the coiled muscle (see Figure S4). The tensile actuation
25
26 is larger than the Joule heating case due to the larger local expansion caused by swelling.
27
28
29
an
30
31
32
dM
33
34
35
36
37
38
39
40
41
42
43
pte

44
45
46
47
48
49
50
51
ce

52
53
54
55
56
Ac

57
58
59 Figure 8: Demonstration of the performance of CF/PDMS artificial muscles. Comparison between the theoretical
60 predictions and the experimental results for the (a) tensile actuation and the (b) specific work of a CF/PDMS muscle
with a radius of 183μm (pentagon symbols) and 400μm (star symbols), coiled using an inserted twist r=8168 turns/m.
Page 17 of 23 AUTHOR SUBMITTED MANUSCRIPT - SMS-106167.R1

1
2
3 Data refer to the condition of the maximum expansion of the PDMS, reached at a temperature of 230°C. (c) A
4 CF/PDMS muscle with a radius of 400μm lifting a half gallon of water more than 1 inch with an applied voltage of
5
0.172 V/cm.
6
7
8 The maximum specific work of 0.758 kJ/kg is very close to the maximum value of 0.836 kJ/kg

pt
9 obtained for hybrid coiled CNTS yarns [3]. Compared to the CNT-based artificial muscles, the main
10
advantages of the CF-based infiltrated muscles are certainly the simple production process and the
11
12 readily available constituents. As explained in the Methods section, a coiled CF/PDMS muscle can
13 be produced from commercial CF without using advanced equipment and with low fabrication

cri
14
15 costs. Producing larger muscles is also straightforward, while being extremely challenging for both
16 CNT coils (due to the production process and the high fabrication cost) and nylon fibers (due to
17
the difficulty in controlling the degree of orientation of the polymeric chains, responsible for the
18
19 anisotropic expansion, during the drawing process [5]). As an example, we demonstrate a
20 CF/PDMS muscle with a radius of 0.4 mm able to lift a half gallon of water more than 1 inch (as

us
21
22 shown in Figure 8(c)), with only 0.172 V/cm (see Supplementary movie 2). An efficiency of ̴1%
23 (calculated as the ratio between the output mechanical energy and the input electrical energy,
24 𝑇 ∆𝑙
25 according to the following equation: 𝜂 = 𝑉 𝐼 𝑡
, where T is the applied tensile load, l the obtained
𝑐
26
27
28
29
an
displacement, V the applied voltage, Ic the applied current, and t the time) is obtained, which is
higher than that one obtained for CNT yarns (0.55%) [3], and close to commercially used shape
memory metals (1-2%) [22].
30
31 The thicker coil is tested for different stress values and results are shown in Figure 8 (a) and (b)
32 (star symbols). It is worth noting how the trend of the tensile actuation and the specific work for
dM
33
34
the bigger muscle is the same of that one related to a fiber with a smaller radius (pentagon
35 symbols), coiled using the same inserted twist r=8168 turns/m. This result confirms that for a
36 certain number of turns, the tensile performances of coiled muscles are scale invariant, as
37
38 observed elsewhere [5, 12].
39 The master curves of Figure 8(a) and (b) are a powerful tool for the design and study of fiber
40
41
reinforced coiled muscles, since they are able to describe their universal performance as a direct
42 consequence of the mechanical properties and the maximum expansion of the guest material,
43
pte

with no dependence on fiber dimensions and number of turns.

44
45
46
47 CONCLUSIONS
48
49 A theoretical model for the tensile actuation of coiled fiber reinforced artificial muscles is
50
51 proposed. This model is able to predict the final configuration of a spontaneously-formed coil
ce

52 obtained by overtwisting a straight elastic fiber, as well as estimate the tensile stroke of the coil
53
when it is thermally or electro-thermally activated.
54
55 The model relates the performances of the muscle to the loading conditions and the thermo-
56 electrical-mechanical properties of the material, allowing the precise tailoring of the output tensile
Ac

57
58 stroke and specific work.
59
60
 AUTHOR SUBMITTED MANUSCRIPT - SMS-106167.R1 Page 18 of 23

1
2
3 Since most of the coiled artificial muscles are composite materials, characterized by fibers that
4
5 constrain a guest material to anisotropic volume expansion, the mechanical anisotropy of the
6 material is also considered for the development of the model.
7
8
The proposed theoretical approach is experimentally validated for produced Carbon Fibers

pt
9 (CF)/Polydimethylsiloxane (PDMS) muscles and an excellent agreement between theoretical
10 predictions and experimental results is observed.
11
12 Furthermore, the CF/PDMS coiled muscles, produced by a simple and low-cost procedure, show
13 promising performance. They are able to lift more than 12,600 times their own weight and provide

cri
14
15
tensile strokes higher than 25%, and a specific work up to 0.758 kJ/kg (more than 18 times higher
16 than the specific work of natural muscles). Big coils, able to lift big loads, are also easy to produce
17 with the analyzed materials.
18
19 The proposed model, reliable and easy to be applied, is a useful tool for the design of fiber
20 reinforced coiled muscles and can encourage future research works toward the enhancements of

us
21
22
such muscles and the development of attractive technologies.
23
24
25 ACKNOWLEDGMENT
26
27
28
29
an
C.L. acknowledges funding from the Beckman Institute Postdoctoral Fellowship at the University of
Illinois Urbana Champaign. S.T. and S.M. acknowledge support from the Mechanical Science and
30 Engineering department at the University of Illinois Urbana Champaign. The authors thank
31 Professor Nancy Sottos for helpful discussions and Matthew Robertson for help with getting
32
dM
33 scanning electron microscope images.
34
35
36
37
38 Appendix - Elastic properties of a twisted fiber reinforced composite
39
40 An analytical model to predict the off-axis elastic properties of a composite characterized by multi-
41 level twisted filaments embedded into a matrix (i.e., the configuration of the straight fiber before
42
43 coiling shown in Figure 3 (c)) was proposed by Naik et al. [18]. Herein we report the principal
pte

44 equations that can be used to relate the off-axis mechanical properties of a twisted impregnated
45
46 yarn to the elastic properties of the fibers and the matrix. More details can be found in the work
47 by Naik et al. [18].
48 The following notation is considered:
49
50  Sij, with i, j = 1, …, 6, are the terms of the compliance matrix of the impregnated straight
51
ce

filament (i.e., a straight fiber embedded into the matrix);

52
53  Sija, with i, j = 1, …, 6, are the terms of the compliance matrix of the twisted impregnated
54 filament (i.e., a twisted fiber embedded into the matrix);
55
56  Sijy, with i, j = 1, …, 6, are the terms of the compliance matrix of the twisted impregnated
Ac

57 yarn (i.e., a yarn of twisted fibers embedded into the matrix);

58 The compliance matrix of the impregnated straight filament [S], is the compliance matrix of an
59
60 orthotropic transversely isotropic material, characterized by five independent elastic constants E1,
E2, ν12, ν23, G12:
Page 19 of 23 AUTHOR SUBMITTED MANUSCRIPT - SMS-106167.R1

1
2
3 1 −𝜈12 −𝜈12
4
0 0 0
𝐸1 𝐸2 𝐸2
5 −𝜈12 1 −𝜈23
6 𝐸1
0 0 0
𝐸2 𝐸2
7 −𝜈12 −𝜈23 1
8 𝐸1 𝐸2
0 0 0
𝐸2
[𝑆] = (A1)

pt
9 2(1+𝜈23 )
10
0 0 0 0 0
𝐸2
11 1
12
0 0 0 0 0
𝐺12
13 1
[ 0 0 0 0 0

cri
14 𝐺12 ]
15
16 The five independent elastic constants can be calculated from the mechanical properties of the
17
18
composite constituents by means of well-known models, such as the rule of mixtures (Equations
19 A2, A3, A4) or Hopkins and Chamis model [23] (Equations A5, A6):
20

us
21 𝐸1 = 𝐸𝑓1 𝑣𝑓 + 𝐸𝑚 𝑣𝑚 (𝐴2)
22
23
24 𝜈12 = 𝜈𝑓12 𝑣𝑓 + 𝜈𝑚 𝑣𝑚 (𝐴3)
25
26 𝜈23 = 𝜈𝑓23 𝑣𝑓 + 𝜈𝑚 𝑣𝑚 (𝐴4)
27
28
29 √𝑣𝑓
an
30 𝐸2 = 𝐸𝑚 [(1 − √𝑣𝑓 ) + 𝐸 ] (𝐴5)
31 1 − √𝑣𝑓 (1 − 𝐸𝑚 )
𝑓2
32
dM
33
34 √𝑣𝑓
35 𝐺12 = 𝐺𝑚 [(1 − √𝑣𝑓 ) + 𝐺 ] (𝐴6)
36 1 − √𝑣𝑓 (1 − 𝐺 𝑚 )
𝑓12
37
38
39 were the subscript m refers to the matrix properties and f to the fiber properties, E is the Young
40 modulus, G the shear modulus, ν the Poisson ratio and v the volumetric fraction.
41
42
Concerning the CF/PDMS composite analyzed in the present work, the volumetric fraction of the
43 carbon fibers vf has been measured by means of SEM analysis on the cross section of the straight
pte

44 fiber (an example is shown in Figure S2(a)) before the coiling procedure (configuration shown in
45
46 Figure 3 (c)).The elastic properties of the PDMS matrix and CF fillers used for the calculations were
47 extracted from the literature [[24-26]].In particular, the following values were used: Em= 3MPa,
48
49
Gm= 1 MPa, νm= 0.5, Ef1= 230 GPa, Ef2= 16 GPa, Gf12= 14 GPa, νf12=0.2, νf23=0.31.
50 The terms of the compliance matrix of the twisted impregnated filament Sija can be related to the
51
ce

terms of the compliance matrix of the impregnated straight filament Sij (listed in Equation A1) as
52
53 follows [18]:
54
𝑎
55 𝑆11 = 𝑆11 𝑐𝑜𝑠 4 𝛼 + 𝑆22 𝑠𝑖𝑛4 𝛼 + (2𝑆12 + 𝑆66 ) 𝑐𝑜𝑠 2 𝛼 𝑠𝑖𝑛2 𝛼
56
Ac

57 1
𝑎
58 𝑆12 = [(𝑆 + 𝑆22 − 𝑆66 ) 𝑐𝑜𝑠 2 𝛼 𝑠𝑖𝑛2 𝛼 + 𝑆12 (𝑐𝑜𝑠 4 𝛼 + 𝑠𝑖𝑛4 𝛼 + 𝑐𝑜𝑠 2 𝛼) + 𝑆23 𝑠𝑖𝑛2 𝛼]
59
2 11
60 𝑎 𝑎
𝑆13 = 𝑆12
 AUTHOR SUBMITTED MANUSCRIPT - SMS-106167.R1 Page 20 of 23

1
2
3 1
𝑎
4 𝑆22 = [3𝑆11 𝑠𝑖𝑛4 𝛼 + (2𝑆12 + 𝑆66 )(1 + 3𝑐𝑜𝑠 2 𝛼)𝑠𝑖𝑛2 𝛼 + 3𝑆22 (1 + 𝑐𝑜𝑠 4 𝛼)
5
8
6 + (2𝑆23 + 𝑆44 )𝑐𝑜𝑠 2 𝛼]
7
𝑎 𝑎
8 𝑆33 = 𝑆22

pt
9
10 𝑎
1
11 𝑆23 = [𝑆 𝑠𝑖𝑛4 𝛼 + 2[𝑆12 𝑐𝑜𝑠 2 𝛼 𝑠𝑖𝑛2 𝛼 + 𝑆22 (1 + 𝑐𝑜𝑠 4 𝛼) − 𝑆44 𝑐𝑜𝑠 2 𝛼 − 𝑆66 𝑠𝑖𝑛4 𝛼 + 6(𝑆12 𝑠𝑖𝑛2 𝛼
8 11
12
13
+ 𝑆23 𝑐𝑜𝑠 2 𝛼)]

cri
14
15 𝑎
1
𝑆44 = [(𝑆 − 2𝑆12 )𝑠𝑖𝑛4 𝛼 + 𝑆22 (1 + 𝑐𝑜𝑠 4 𝛼) + (𝑆44 − 2𝑆23 )𝑐𝑜𝑠 2 𝛼 + 𝑆66 𝑠𝑖𝑛2 𝛼(1 + 𝑐𝑜𝑠 2 𝛼)]
16 2 11
17
18 𝑎
1
19 𝑆55 = 2(𝑆11 − 2𝑆12 + 𝑆22 )𝑐𝑜𝑠 2 𝛼 𝑠𝑖𝑛2 𝛼 + (𝑆66 [𝑐𝑜𝑠 2 (2𝛼) + 𝑐𝑜𝑠 2 𝛼] + 𝑆44 𝑠𝑖𝑛2 𝛼)
2
20

us
21 𝑎
𝑆66 𝑎
= 𝑆55
22
23
(𝐴7)
24
25
26 were α is the twist angle that can be optically measured (Figure S2 (a)).
27
28
29
be obtained by the following relationships [18]:
an
Known the matrix [Sa], the terms of the compliance matrix of the twisted impregnated yarn Sijy can

30 𝑦 𝑎 𝑎 𝑎 𝑎
31 𝑆11 = 𝑆11 𝐼1 + 𝑆22 𝐼2 + (2𝑆11 + 𝑆66 )𝐼4
32
dM
33 𝑦 1
34 𝑆12 = [(𝑆 𝑎 + 𝑆22
𝑎 𝑎
− 𝑆66 𝑎 (𝐼
)𝐼4 + 𝑆12 𝑎
1 + 𝐼2 + 𝐼3 ) + 𝑆23 𝐼5 ]
2 11
35
36 𝑦 𝑦
𝑆13 = 𝑆12
37
38 1
𝑦 𝑎 𝑎 𝑎 )(𝐼 𝑎 𝑎
39 𝑆22 = [3𝑆11 𝐼2 + (2𝑆12 + 𝑆66 𝑎
5 + 3𝐼4 ) + 3𝑆22 (1 + 𝐼1 ) + (2𝑆23 + 𝑆44 )𝐼3 ]
40 8
41
42 𝑦 1 𝑎 𝑎 𝑎 (𝐼 𝑎 𝑎 𝑎 𝑎
𝑆23= [3𝑆11 𝐼2 + 2𝑆12 𝐼4 + 𝑆22 6 + 𝐼1 ) − 𝑆44 𝐼3 − 𝑆66 𝐼2 + 6(𝑆12 𝐼5 + 𝑆23 𝐼3 )]
43 8
pte

44
𝑦 𝑦
45 𝑆33 = 𝑆22
46
47 1
𝑦 𝑎 𝑎 𝑎 )𝐼 𝑎 𝑎
48 𝑆55 = 2(𝑆11 − 2𝑆12 + 𝑆22 4 + [(𝑆44 𝐼5 + 𝑆66 (𝐼1 + 𝐼2 + 𝐼3 − 𝐼4 )]
49 2
50 𝑦 𝑦
51 𝑆66 = 𝑆55
ce

52
53 𝑦 1
54
𝑆44 = [(𝑆 𝑎 − 2𝑆12
𝑎 )𝐼 𝑎 𝑎 𝑎 𝑎
2 + 𝑆22 (𝐼1 + 𝐼6 ) + (𝑆44 − 2𝑆23 )𝐼3 + 𝑆66 (𝐼5 + 𝐼4 )]
2 11
55
56 (𝐴8)
Ac

57
58 were the coefficients I1 to I6 can be calculated as:
59
60
Page 21 of 23 AUTHOR SUBMITTED MANUSCRIPT - SMS-106167.R1

1
2
3 1
4 𝐼1 = (𝑐𝑜𝑡 2 𝛼)(1 − 𝑐𝑜𝑠 2 𝛼 + 𝑠𝑖𝑛2 𝛼)
5
2
6 1
7 1
8 𝐼2 = (2𝑐𝑜𝑡 2 𝛼) ( 22 + 2𝑙𝑛𝑐𝑜𝑠𝛼 − 𝑐𝑜𝑠 2 𝛼)
𝑐𝑜𝑠 𝛼 2

pt
9
10
11
𝐼3 = (2𝑐𝑜𝑡 2 𝛼)𝑙𝑛𝑠𝑒𝑐 𝛼
12
13
𝑐𝑜𝑠 2 𝛼 1

cri
14 2
𝐼4 = (2𝑐𝑜𝑡 𝛼) ( − 𝑙𝑛𝑐𝑜𝑠𝛼 − )
15 2 2
16
17 1
18 1
19 𝐼5 = (2𝑐𝑜𝑡 𝛼) (𝑙𝑛𝑐𝑜𝑠𝛼 + 22 − )
2
𝑐𝑜𝑠 𝛼 2
20

us
21
22 1
23 2 𝑠𝑖𝑛2 𝛼
24 𝐼6 = (2𝑐𝑜𝑡 2 𝛼) ( )
𝑐𝑜𝑠 2 𝛼
25
26
27
28
29
an (𝐴9)

The off-axis elastic properties of the impregnated yarn, that corresponds to the configuration of
30
31 the straight fiber before coiling shown in Figure 3 (c), can be determined as follows:
32
dM
33 1 1 1
34 𝐸𝑥 =𝑦, 𝐸𝑦 = 𝑦 , 𝐸𝑧 = 𝑦 ,
𝑆11 𝑆22 𝑆33
35
36 1 1 1
37
𝐺𝑦𝑧 = 𝑦, 𝐺𝑥𝑧 = 𝑦, 𝐺𝑥𝑦 = 𝑦,
𝑆44 𝑆55 𝑆66
38 𝑦 𝑦 𝑦
39 𝑆 𝑆 𝑆
40
𝜈𝑦𝑧 = − 23
𝑦 , 𝜈𝑧𝑥 = − 13
𝑦 , 𝜈𝑥𝑦 = − 12
𝑦
𝑆22 𝑆22 𝑆11
41
42
43
(𝐴10)
pte

44
45 The values of Ex and Gyz to use in Equations 20-23 of the manuscript can then be calculated from
46 Equation A10.
47
48
49
50
51
ce

52
53
54
55
56
Ac

57
58
59
60
 AUTHOR SUBMITTED MANUSCRIPT - SMS-106167.R1 Page 22 of 23

1
2
3 REFERENCES
4
5
[1] Haines C S, Li N, Spinks G M, Aliev A E, Di J T and Baughman R H 2016 New twist on artificial
6
7
muscles Proceedings of the National Academy of Sciences of the United States of America 113
8 11709-16

pt
9 [2] Foroughi J, Spinks G M, Wallace G G, Oh J, Kozlov M E, Fang S L, Mirfakhrai T, Madden J D W, Shin
10 M K, Kim S J and Baughman R H 2011 Torsional Carbon Nanotube Artificial Muscles Science 334
11 494-7
12 [3] Lima M D, Li N, de Andrade M J, Fang S L, Oh J, Spinks G M, Kozlov M E, Haines C S, Suh D, Foroughi
13 J, Kim S J, Chen Y S, Ware T, Shin M K, Machado L D, Fonseca A F, Madden J D W, Voit W E, Galvao D

cri
14 S and Baughman R H 2012 Electrically, Chemically, and Photonically Powered Torsional and Tensile
15
Actuation of Hybrid Carbon Nanotube Yarn Muscles Science 338 928-32
16
17 [4] Lima M D, Hussain M W, Spinks G M, Naficy S, Hagenasr D, Bykova J S, Tolly D and Baughman R H
18 2015 Efficient, Absorption-Powered Artificial Muscles Based on Carbon Nanotube Hybrid Yarns
19 Small 11 3113-8
20 [5] Haines C S, Lima M D, Li N, Spinks G M, Foroughi J, Madden J D W, Kim S H, Fang S L, de Andrade M

us
21 J, Goktepe F, Goktepe O, Mirvakili S M, Naficy S, Lepro X, Oh J Y, Kozlov M E, Kim S J, Xu X R,
22 Swedlove B J, Wallace G G and Baughman R H 2014 Artificial Muscles from Fishing Line and Sewing
23 Thread Science 343 868-72
24
[6] Mirvakili S M, Pazukha A, Sikkema W, Sinclair C W, Spinks G M, Baughman R H and Madden J D W
25
26
2013 Niobium Nanowire Yarns and their Application as Artificial Muscles Advanced Functional
27
28
29
[7]
Materials 23 4311-6
an
Cheng H H, Hu Y, Zhao F, Dong Z L, Wang Y H, Chen N, Zhang Z P and Qu L T 2014 Moisture-
Activated Torsional Graphene-Fiber Motor Advanced Materials 26 2909-13
30 [8] Lee J, Ko S, Kwon C H, Lima M D, Baughman R H and Kim S J 2016 Carbon Nanotube Yarn-Based
31 Glucose Sensing Artificial Muscle Small 12 2085-91
32 [9] Yip M C, Niemeyer G and Ieee 2015 High-Performance Robotic Muscles from Conductive Nylon
dM
33
Sewing Thread. In: IEEE International Conference on Robotics and Automation (ICRA), (Seattle, WA
34
pp 2313-8
35
36 [10] Kim S H, Haines C S, Li N, Kim K J, Mun T J, Choi C, Di J T, Oh Y J, Oviedo J P, Bykova J, Fang S L, Jiang
37 N, Liu Z F, Wang R, Kumar P, Qiao R, Priya S, Cho K, Kim M, Lucas M S, Drummy L F, Maruyama B,
38 Lee D Y, Lepro X, Gao E L, Albarq D, Ovalle-Robles R, Kim S J and Baughman R H 2017 Harvesting
39 electrical energy from carbon nanotube yarn twist Science 357 773-8
40 [11] Love A E H 1944 A treatise on the Mathematical Theory of Elasticity (New York: Dover Publications)
41 [12] Aziz S, Foroughi J, Brown H R and Spinks G M 2016 Controlled and Scalable Torsional Actuation of
42 Twisted Nylon 6 Fiber Journal of Polymer Science Part B-Polymer Physics 54 1278-86
43
pte

[13] Thompson J M T and Champneys A R 1996 From helix to localized writhing in the torsional post-
44
45 buckling of elastic rods Proceedings of the Royal Society a-Mathematical Physical and Engineering
46 Sciences 452 117-38
47 [14] Timoshenko SP G J 1961 Theory of elastic stability (New York: McGraw-Hill)
48 [15] 2017 Structure and Properties of High-Performance Fibers Structure and Properties of High-
49 Performance Fibers 187 1-417
50 [16] Choy C L, Chen F C and Young K 1981 NEGATIVE THERMAL-EXPANSION IN ORIENTED CRYSTALLINE
51
ce

POLYMERS Journal of Polymer Science Part B-Polymer Physics 19 335-52

52
[17] Chun K Y, Kim S H, Shin M K, Kwon C H, Park J, Kim Y T, Spinks G M, Lima M D, Haines C S,
53
54
Baughman R H and Kim S J 2014 Hybrid carbon nanotube yarn artificial muscle inspired by spider
55 dragline silk Nature Communications 5
56 [18] Naik N K and Madhavan V 2000 Twisted impregnated yarns: elastic properties Journal of Strain
Ac

57 Analysis for Engineering Design 35 83-91

58 [19] Kim B, Park M, Kim Y S and Jeong U 2011 Thermal Expansion and Contraction of an Elastomer
59 Stamp Causes Position-Dependent Polymer Patterns in Capillary Force Lithography Acs Applied
60 Materials & Interfaces 3 4695-702
[20] Zill D and Cullen C 1997 Differential Equations with Boundary-Value Problems (Pacific Grove, CA
Page 23 of 23 AUTHOR SUBMITTED MANUSCRIPT - SMS-106167.R1

1
2
3 [21] Liu M, Sun J R and Chen Q F 2009 Influences of heating temperature on mechanical properties of
4 polydimethylsiloxane Sensors and Actuators a-Physical 151 42-5
5
[22] Huber J E, Fleck N A and Ashby M F 1997 The selection of mechanical actuators based on
6
7
performance indices Proceedings of the Royal Society a-Mathematical Physical and Engineering
8 Sciences 453 2185-205

pt
9 [23] DA H 1985 A unique set of micromechanics equations for high-temperature metal
10
11 matrix composites. ed C C (Nashville: NASA TM 87154)
12 [24] Johnston I D, McCluskey D K, Tan C K L and Tracey M C 2014 Mechanical characterization of bulk
13 Sylgard 184 for microfluidics and microengineering Journal of Micromechanics and

cri
14 Microengineering 24
15 [25] Mounier D e a Evaluation of transverse elastic properties of fibers used in composite materials by
16 laser resonant ultrasound spectroscopy. Proceedings of the Acoustic 2012 Nantes Conference)
17
[26] Krucinska I and Stypka T 1991 DIRECT MEASUREMENT OF THE AXIAL POISSON RATIO OF SINGLE
18
19 CARBON-FIBERS Composites Science and Technology 41 1-12
20

us
21
22
23
24
25
26
27
28
29
an
30
31
32
dM
33
34
35
36
37
38
39
40
41
42
43
pte

44
45
46
47
48
49
50
51
ce

52
53
54
55
56
Ac

57
58
59
60