Medical Engineering and Physics 48 (2017) 55–61

Contents lists available at ScienceDirect

Medical Engineering and Physics

journal homepage: www.elsevier.com/locate/medengphy

Thread-based microfluidics: Flow patterns in homogeneous and

heterogeneous microfiber bundles
J. Berthier a,b,∗, K.A. Brakke c, D. Gosselin a,b, E. Berthier d,e, F. Navarro a,b
a
Univ. Grenoble Alpes, F-38000 Grenoble, France
b
CEA LETI, MlNATEC Campus, F-38054 Grenoble, France
c
Mathematics Department, Susquehanna University, Selinsgrove, PA 17870, USA
d
Tasso Inc., 1631 15th Ave. W, Seattle, WA 98119, USA
e
Department of Chemistry, University of Washington, Seattle, WA 98105, USA

a r t i c l e i n f o a b s t r a c t

Article history: Thread-based microfluidics has recently seen considerable developments in the domain of portable di-
Received 14 December 2016 agnostic systems, smart bandages and tissue engineering. Similarly to paper-based microfluidics, thread-
Revised 1 June 2017
based microfluidics uses the wicking of fibers to move fluids. It has the advantage of confining and guid-
Accepted 9 August 2017
ing the fluid along the yarns in a one, two or three dimensional space.
A global approach to the motion of fluids in yarns and fiber bundles has already been reported in
Keywords: the literature based on the Lucas–Washburn–Rideal law. However no detailed investigation of the flow
Thread-based microfluidics pattern inside the bundle has been conducted, depending on the internal structure of the bundle. Espe-
Heterogeneous yarns cially when the bundle possesses heterogeneous wetting properties, such as two different wetting regions
Liquid caging
interior and exterior, different flow patterns may exist.
In this work, we perform a theoretical and numerical analysis of the different flow regimes for ho-
mogenous and heterogeneous fiber bundles. It is demonstrated that a limited number of fibers is suf-
ficient for thread-based capillary flows, and that a caging of the flow can be achieved by realizing a
lyophobic envelope.
© 2017 IPEM. Published by Elsevier Ltd. All rights reserved.

1. Introduction technologies have been developed using woven yarns [4–7].

Thread-based systems and biosensors have five main advantages
Thread-based microfluidics (TBM) is a relatively new concept over paper-based systems: first, the flexibility of threads make
of microfluidics which uses the wicking of fiber bundles and wo- them good candidates for the creation of three-dimensional (3D)
ven yarns to move fluids. This concept has recently seen impor- microfluidic circuits. Thread-based systems can be one, two or
tant developments in the domain of portable diagnostic systems, three dimensional, while paper-based systems are mostly two
smart bandages and tissue engineering. Usually, the threads con- dimensional, two and a half dimensional at best [8].
sist of fibers of natural origin such as cotton or cellulose, or of ar- The second advantage over paper-based systems is the align-
tificial origin such as polymers, aligned or twisted together (Fig. 1). ment of the fibers, which facilitates the circulation of the liquids
Wetting properties of the fibers are necessary to obtain wicking. In (Fig. 2).
the case of biologic sample liquids which are aqueous, either the Third, the confinement of fluids is better in yarns, and a lesser
fibers are naturally hydrophilic—like cellulose—or they must be hy- volume of fluid is lost unused in the fibers. Fourth, yarns can be
drophilically treated in order to obtain a capillary flow [1]. For ex- made of different types of natural fibers such as cotton, silk, wool,
ample, wool fibers can be treated by chlorination [1], most natural etc., which allows for easy recycling.
fibers by introduction of silane coupling agents [2], and polyester Finally, as we shall see in this work, threads of different materi-
thread can be treated by plasma oxidation [3]. als or surface treatment can be assembled easily to precisely guide
In a manner similar to paper-based strips, new concepts and the fluid flow. The same approach is much more complicated for
applications of low-cost, portable, and field-based diagnostic paper-based microsystems.
Biosensors using microfluidic thread technology have recently
∗
been developed: for example passive blood-plasma separation
Corresponding author at: CEA LETI, MlNATEC Campus, F-38054 Grenoble,
using fibers coated with immoglobulin [9] and potassium ar-
France.
E-mail addresses: [email protected], [email protected] (J. Berthier). rays based on ionophore extraction chemistry for the optical

http://dx.doi.org/10.1016/j.medengphy.2017.08.004
1350-4533/© 2017 IPEM. Published by Elsevier Ltd. All rights reserved.
56 J. Berthier et al. / Medical Engineering and Physics 48 (2017) 55–61

Fig. 3. Homogeneous and heterogeneous fiber bundles: liquid in green color, hy-
drophilic (lyophilic) fibers in yellow color, hydrophobic (or less hydrophilic) fibers
in brown color. (For interpretation of the references to color in this figure legend,
the reader is referred to the web version of this article.)

Fig. 1. Different types of yarns: A and B: cotton yarns; C: plasma O2 treated poly-
meric fibers.

Fig. 2. Different types of fiber matrix: A, random matrix corresponding to paper; B,

aligned fibers found in yarns.

recognition of potassium [10]. In medicine, smart bandages have

been developed for the monitoring of the healing of skin wounds
[11,12], or for surgical recovery [13]. These bandages use TBM
to transport and guide body fluids toward detectors. Also, tissue
Fig. 4. SCF between two parallel rods: liquid is colored green, the contact angle is
engineering makes a wide use of TBM to achieve the seeding and 30°. (For interpretation of the references to color in this figure legend, the reader is
immobilization of cells which are left to proliferate on the surface referred to the web version of this article.)
of the woven substrate [14,15].
Wicking of threads has been investigated in the literature, ei-
ther in the case of solitary fibers [16], or from a global, average 2. SCF theory
standpoint [17–18]. At the present time, thread-based microflows
have been investigated for homogeneous yarns, i.e. yarns made A universal law has been derived from the expression of the
of similar fibers, with the same wetting properties. Capillary flow Gibbs free energy [24] that predicts if a spontaneous capillary flow
behavior in heterogeneous yarns has not been studied so far. In (SCF) occurs in open or closed configurations for a uniaxial flowing
this work, we show that different flow patterns exist in heteroge- liquid. This law states that spontaneous capillary flow occurs if the
neous fiber bundles. We investigate theoretically the capillary flow generalized Cassie equation is satisfied [20]
patterns in homogeneous and heterogeneous fiber bundles using

recent developments of open capillarity [19–21] and numerically pW,i cos θi + pF cos π > 0, (1)
with the software Surface Evolver (Fig. 3) [22]. The heterogeneous i
bundles considered in this paper have an outer layer of fibers with
different wetting properties than the inner fibers [23]. where the notations pW ,i correspond to the wetted parts of the
We show that depending on the contact angles of the liq- perimeter in a cross section, pF is the free liquid perimeter—the
uid with the outer and inner fibers, and on the compactness of air–liquid boundary—and θ i are the contact angles. For example,
the bundle, different flow patterns exist, i.e. different regions are Fig. 4 shows a SCF between two parallel rods. In this particular
wetted or not. Especially, we determine the conditions for which case, relation (1) can be written
only the interior part of the fiber bundle is wetted, by realizing a
lyophobic envelope of the bundle. ( pW 1 + pW 2 ) cos θ > ( pF 1 + pF 2 ). (2)
 J. Berthier et al. / Medical Engineering and Physics 48 (2017) 55–61 57

Fig. 6. Cross section of the bundle; front view of the liquid in the bundle: C is the
distance between rod centers, R is the rod radius, and θ is the contact angle of the
liquid on the rods.

satisfied. At least two parallel threads are required to obtain a SCF

(Fig. 4).
Note that the area of the cross section is

A ≈ π Rout 2 , (6)
and the area of the fibers

AF = n π R2 . (7)
The fraction τ = A/AF in the compact bundle depends on the ar-
rangement of the fibers, but in any case this fraction is approxi-
Fig. 5. Top: cross section of a compact homogeneous fiber bundle; bottom: SCF mately comprised between
domain, the two curves correspond respectively to τ = 1 (green curve) and τ = 2
(orange curve). (For interpretation of the references to color in this figure legend, 1 < τ < 2. (8)
the reader is referred to the web version of this article.)
The lower bound stems from the definition of A and AF . The
upper bound τ = 2 is linked to the compaction of the fiber bundle
In the following, we use relation (1) to analyze SCFs in fiber where each thread touches its neighbors. Using (6) and (7), we find
bundles, starting with the case of homogeneous microfiber bun-
dles. π Rout 2 = n π R2 τ . (9)
3. Homogeneous fiber bundle Then, rearranging (9)
Rout √
In a homogeneous fiber bundle, all the fibers have the same = n τ. (10)
wettability. Fiber arrangements are numerous. Let us first consider R
a compact bundle (Fig. 5). The free perimeter pF is approximated Substitution of (10) in (5) yields
by 
τ
pF ≈ 2 π Rout . (3) < cos θ , with 1 < τ < 2, (11)
n
On the other hand, assuming that all the fibers are totally Relation (11) is plotted in Fig. 5. The figure shows an asymptote
wetted—which is true except for some exterior fibers—the wetted just above 80°. It shows that it is not necessary to twist together
perimeter is simply approximated by more than 30 fibers in a bundle: Young contact angles up to 75°
pW ≈ n 2 π R, (4) will produce a SCF in such bundles. This observation of a number
of approximately 25–30 threads is in agreement with the threads
where n is the number of fibers in the bundle. Upon substitution of used by Ballerini and Wei in their experiments (Fig. 1C) [16].
(3) and (4) in (1) we derive the condition for SCF in a fiber bundle A more detailed approach can be done. Let us denote by N the
Rout number of rods at the periphery of the bundle, and by M the num-
< cos θ . (5) ber of rods in the interior of the bundle. With the notations of
nR
Fig. 6, the free perimeter pF is given by
This inequality shows that a single thread cannot induce a SCF:
in such a case, n = 1, and because Rout > R, condition (5) is never pF = N (C − 2 R sin θ ), (12)
58 J. Berthier et al. / Medical Engineering and Physics 48 (2017) 55–61

Fig. 7. A: SCF in a relatively spaced bundle (C/R = 3); B: SCF in a compact bundle
(C/R = 2.2); C: spaced bundle, SCF limit for C/R = 5.2. (A, B, C: contact angle θ = 50°).
D: SCF domain for the homogeneous square bundle: the theoretical limit is the or- Fig. 8. Evolver image of a heterogeneous bundle of parallel rods wicked by water
ange line, the green dots are Evolver results. (For interpretation of the references to (water-solid contact angle in the inner ring 40°, water-solid contact angle in the
color in this figure legend, the reader is referred to the web version of this article.) outer ring 50°, distance between rods centers is three times the rod radius, C = 3
R).

regardless of the shape of the bundle, as long as all the peripheral

bundles are too far from each other, the situation is that of single
rod centers are exactly C apart, and the wetted perimeter pW is
rods, and no SCF can be achieved.
the sum of the wetted perimeter on the interior rods, plus the full
A comparison between theoretical and numerical approaches is
perimeter the outer rods minus the outer part of the four corners
shown in Fig. 7D, showing the agreement between the two ap-
and minus the outer not-wetted arcs:
proaches.
pw = 2 π R M + (2 π R N − 2 π R − 2 θ R N ). (13)
4. Heterogeneous fiber bundle
Substitution of (12) and (13) in (1) yields the SCF condition for
a homogeneous bundle:
C  We consider now a heterogeneous bundle in which the outer
N − 2 sin θ rods (outer ring) have different wetting properties than the inner
< cos θ .
R
(14) rods. We show here that different flow patterns exist depending
2 π (N + M ) − 2 π − 2 θ N
on the two contact angles (internal and external) and compactness
Note that the dimensions C and R only appear through the di- of the bundle, characterized by the length scale C/R.
mensionless parameter C/R. The contact angle θ lim at the SCF limit The approach consists in running Evolver for contact angles be-
is then the solution of the equation tween 0 and 2π and compactness ratios larger than 2 (ratio for
A cos θlim + B sin θlim = D, (15) which the fibers are in contact). A flow regime or flow pattern is
by definition defined by its contour in a cross section of the bun-
where dle, i.e. by its free and wetted perimeters. It is checked that the
A = 2 π (N + M ) − 2 π − 2 θlim N, threshold between each flow pattern agrees with that determined
by relation (1) using the proper geometry of the flow contour. For
B = 2 N, (16)
simplicity, we do not give the details of each geometry here, but
C
D=N R
. we present only the flow phase diagrams.
A numerical approach to the SCF onset in microfiber bundle can
be done using the numerical program Surface Evolver [22]. Let us 4.1. Square bundle
recall that Surface Evolver does not treat the dynamics of the flow,
but iteratively relocates the interface in the direction of lower en- We consider a square bundle, geometrically identical to that of
ergy. In our case, it is the stop or start of the flow that we inves- Fig. 6, but where the outer rods have different wetting properties
tigate and no dynamics is involved. Hence Evolver can be used to than the inner rods (Fig. 8). Different flow patterns are found de-
predict the SCF condition. Moreover, in the case of a slow, iner- pending on the three parameters θ 1 and θ 2 , respectively, the con-
tialess motion, it has been checked that Evolver produces accurate tact angles of the liquid with the inner and outer rods, and the
results. This occurs if the Reynolds, Weber and capillary numbers compactness ratio C/R. In a general manner, the SCF condition de-
are all very small, which is frequent for capillary microflows [25]. rived from (1) is
Fig. 7 shows the SCF in a square bundle for different compactness.
pW 1 cos θ1 + pW 2 cos θ2 + pF cos π > 0, (17)
Compactness of the bundle is characterized by the ratio C/R where
C is the distance between rod centers and R is the radius of the where the indices 1, 2 and F stand for inner ring, outer ring and
rod. (free) air. The determination of the different flow patterns requires
Note that the SCF depends on the contact angle and the spacing long geometrical developments. In order to indicate the approach
of the rods. For example the maximum spacing for SCF for a con- followed, we derive the condition for the case where all the rods
tact angle θ = 50° is C/R = 5.2, as shown in Fig. 7C. Clearly, if the are in contact with the liquid, such a case we named “all outer”
 J. Berthier et al. / Medical Engineering and Physics 48 (2017) 55–61 59

Fig. 9. The different flow patterns in a square bundle depending on the compact-
ness of the bundle and contact angles.

Fig. 11. Phase diagram in a square bundle of 16 cylindrical fibers with different
wettabilities of the outer ring and inner ring (C/R = 2.1). The phase diagram in-
dicates 7 regions: no SCF (yellow), bulk SCF (dark green), interior-only SCF with
dewetted outer rods (light green), bulk SCF with dewetted corners (violet and light
violet), exterior-only SCF with retreat in the middle (orange and dark orange). (For
interpretation of the references to color in this figure legend, the reader is referred
to the web version of this article.)

Finally, the free perimeter with air is

pF = N (C − 2 R sin θ2 ). (20)

Using (17), the SCF condition for this flow configuration is

  
M 1 θ2
2π cos θ1 + 2 π 1− − cos θ2
N N π (21)
C
> − 2 sin θ2 .
R
The different patterns observed depending on the rod wettabil-
ity and compactness ratio have been found to be: all wetted (case
A in Fig. 9) where the liquid contacts all rods, center only (case B in
Fig. 9) where the liquid does not make contact with the outer rods,
all but corners and wonky corners (case C in Fig. 9) where the cor-
ner rods are completely dewetted (and the corner inner rods are
Fig. 10. Phase diagram in a square bundle of 16 cylindrical fibers with different or are not dewetted), all outer (case D in Fig. 9) where the flow is
wettabilities of the outer ring and inner ring (C/R = 3). The phase diagram indicates located between the outer rods, with contact with the neighboring
5 regions: no SCF (yellow), bulk SCF (dark green), interior-only SCF with dewetted rods inside, outer ring (case E in Fig. 9) where the liquid wets the
outer rods (light green), bulk SCF with dewetted corners (violet), exterior-only SCF outer ring and not the inner rods.
with retreat in the middle (orange). (For interpretation of the references to color in
this figure legend, the reader is referred to the web version of this article.)
For a given compactness ratio C/R, flow phase diagrams can be
plotted as functions of the two contact angles, showing the differ-
ent flow domains. Figs. 10 and 11 show the flow phase diagram
for compactness ratio C/R = 3 and C/R = 2.1. The diagrams are simi-
(Fig. 8). The wetted perimeter in the inner ring pW 1 is
lar, except for the “wonky corners” and “outer ring” morphologies
that only appear for the most compact bundles.
pW 1 = 2 π R M. (18)
In particular, the condition (21) between the contact angles θ 1
The wetted perimeter along the outer rods is the total area of and θ 2 corresponds to the limit between “all outer” domain and
the outer rods length, minus the outer part of the four corners and “no SCF” domain in Fig. 10.
minus the outer not-wetted arcs along the rods. The wetted area
along the outer rods is then 4.2. Round bundle

pW 2 = 2 π R N − (2 π R + 2 θ2 R N ) A similar approach can be conducted for round bundles. Four

  (19)
= 2π RN 1 − 1
N
− θπ2 . flow patterns are found: all wetted (Fig. 12A), center only (Fig. 12B),
60 J. Berthier et al. / Medical Engineering and Physics 48 (2017) 55–61

Fig. 13. A: optical fiber impregnated by elastomer, reprinted from the patent [27];
B: caging of fibers obtained by using a hydrophobic (lyophobic) outer ring forming
a hydropobic envelope.

Fig. 14. A: whole blood flowing in a parallel fiber bundle, the blood cells are in
green color; B: sketch of fibers wicking observed by Nilghaz et al. [9]. (For inter-
pretation of the references to color in this figure legend, the reader is referred to
the web version of this article.)

Fig. 12. Top: Different flow patterns in the case of a round bundle: A, all wetted; B:
center only; C, all outer; D, outer ring. Bottom: phase diagram for a rounded bundle together. This work shows that a “macro” yarn of adjustable wetta-
as a function of the inner and outer contact angles.
bility can be obtained comprising sub-yarns of different wettabili-
ties.
all outer (Fig. 12C) and outer ring (Fig. 12D). Because no bundle cor-
ners exist in the case of a round bundle, the all but corners and Conflict of interest
wonky corners flow regimes do not appear.
No conflict of interest.
5. Discussion and conclusion
Acknowledgments
In this work, we have demonstrated that different flow patterns
This work was funded by the CEA-Leti, in collaboration with
exist in heterogeneous fiber bundles, depending on the contact an-
Susquehanna University, University of Washington and Tasso Inc.
gles and compactness of the bundle. Applications can be foreseen
in the domains of optics, smart textiles, space, biotechnology and References
biomedicine. Smart bandages could gain from being heterogeneous
in some cases, as for example when using conductive yarns with [1] Ferrero F, Periolatto M. Modification of surface energy and wetting of textile
an insulating wrapping [26]. fibers. Wetting and wettability. Aliofkhazraei M, editor. INTECH; 2015. Chapter
6.
In optics, impregnation of fibers is used for the fabrication of [2] Adekunle KF. Surface treatments of natural fibers, review, part 1. Open J Polym
optical fibers [27–29]: in this case, the glass fiber bundle is sta- Chem 2015;5:41–6.
bilized by an elastomer, which is liquid at first, then reticulated [3] Ballerini DR, Li X, Shen W. Flow control concepts for thread-based microfluidic
devices. Biomicrofluidics 2011;5:014105.
(Fig. 13A). This work shows that a heterogeneous bundle can be [4] Safavieh R, Mirzaei M, Qasaimeh MA, Juncker D. Yarn-based microfluidics:
used for impregnation problems by caging—i.e. building a hy- from basic elements to complex circuits. In: Thirteen International Conference
drophobic envelope—of the capillary flow (impregnating flow) in- on Miniaturized Systems for Chemistry and Life Sciences, Jeju, Korea; 2009
November 1–5.
side the bundle (Fig. 13B). [5] Li X, Tian J, Shen W. Thread as a versatile material for low-cost microfluidic
The principle pointed out in this work can be of inter- diagnostics. ACS Appl Mater Interfaces 2010;2(1):1–6.
est in biotechnology. For example, in biotechnology and biology, [6] Reches M, Mirica KA, Dasgupta R, Dickey MD, Butte MJ, Whitesides GM. Thread
as a matrix for biomedical assays. ACS Appl Mater Interfaces 2010;2(6):1722–8.
very recent experiments using threads to separate plasma from
[7] Nilghaz A, Ballerini DR, Shen W. Exploration of microfluidic devices
whole blood are showing some similarities with our approach [8] based on multi-filament threads and textiles: a review. Biomicrofluidics
(Fig. 14). A hydrophobic caging of the capillary flow would prevent 2013;7(5):051501.
[8] Martinez AW, Phillips ST, Whitesides GM. Three-dimensional microfluidic de-
contamination. This is a general case in bioanalysis (μTAD, micro
vices fabricated in layered paper and tape. PNAS 2008;105:19606–11.
thread analysis devices), when potentially contaminated samples [9] Nilghaz A, Ballerini DR, Guan L, Li L, Shen W. Red blood cell transport mech-
of human fluids are used. anisms in polyester thread-based blood typing devices. Anal Bioanal Chem
Finally, one may remark that heterogeneous bundles with for 2016;408(5):1365–71.
[10] Erenas MM, de Orbe-Payá I, Capitan-Vallvey LF. Surface modified thread-based
example, a hydrophobic outer ring and a hydrophilic interior, are microfluidic analytical device for selective potassium analysis. Anal Chem
not easy to fabricate. But the bundles or yarns can be assembled 2016;88:5331–7.
 J. Berthier et al. / Medical Engineering and Physics 48 (2017) 55–61 61

[11] Sharp D. The development of smart-bandage technologies Thesis at. Notting- [21] Berthier J, Brakke KA, Berthier E. Open microfluidics. Scrivener-Wiley Publish-
ham Trent University; 2009. ing; 2016.
[12] https://www.csiro.au/en/Research/MF/Areas/Chemicals- and- fibres/Advanced- [22] Brakke KA. The surface evolver. Exp Math 1992;1(2):141–65.
fibres/Smart- clothing- and- textiles/Smart- bandages- reveal- healing/: Smart ban- [23] Brakke KA, Gosselin D, Berthier E, Berthier J. Thread-based microfluidics: spon-
dages reveal healing. taneous capillary flow in homogeneous and heterogeneous microfiber bundles.
[13] Mostafalu P, Akbari M, Alberti KA, Xu Q, Khademhosseini A, Sonkusale SR. In: Proceedings of the 2016 Micro and Nano Flows Conference, Milano; 2016
A toolkit of thread-based microfluidics, sensors, and electronics for 3D tissue September 11–14.
embedding for medical diagnostics. Microsyst Nanoeng 2016;2:16039. [24] Gibbs JW. A method of geometrical representation of the thermodynamic
[14] Young CD, Wu JR, Tsou TL. High-strength, ultra-thin and fiber-reinforced properties of substances by means of surfaces. Trans Connecticut Acad Arts
pHEMA artificial skin. Biomaterials 1998;19(19):1745–52. Sci 1873;2:382–404.
[15] Chirila TV. An overview of the development of artificial corneas with [25] Berthier J, Brakke KA, Gosselin D, Navarro F, Belgacem N, Chaussy D. Spon-
porous skirts and the use of PHEMA for such an application. Biomaterials taneous capillary flow in curved, open microchannels. Microfluid Nanofluid
2001;22(24):3311–17. 2016;20:100.
[16] Quéré D. Fluid coating on a fiber. Annu Rev Fluid Mech 1999;31:347–84. [26] Cherenack K, van Pieterson L. Smart textiles: challenges and opportunities. J
[17] Trybala A, Bureiko A, Kovalchuk N, Arjmandi-Tash O, Liu Z, Starov V. Wetting Appl Phys 2012;112:091301.
properties of cosmetic polymeric solutions on hair tresses. Colloids Interface [27] N.S. Janetos, A. Marzocchi, Glass fiber bundle comprising an elastomer com-
Sci Commun 2015;9:12–15. patible impregnant and a thickening agent, US patent 3705073 A, 1972.
[18] Tsai C-C, Gu Y, Kornev KG. Wetting of nanofiber yarns. Colloids Surf A: Physic- [28] A. Marzocchi, Glass fiber reinforced elastomers, US patent 3765927 A, 1973.
ochem Eng Aspects 2014;459:22–30. [29] Wang R-M, Zheng S-R, Zheng YG. Polymer matrix composites and technology.
[19] Berthier J, Brakke KA. The physics of microdroplets. Scrivener-Wiley Publish- Woodhead Publishing in Materials, Elsevier; 2011.
ing; 2012.
[20] Berthier J, Brakke K, Berthier E. A general condition for spontaneous cap-
illary flow in uniform cross-section microchannels. Microfluid Nanofluid
2014;16(4):779–85.