Liquid marble is a recently emerging digital microfluidic platform with a wide range of applications. Conventional liquid marbles are synthesized by coating liquid droplets with a thin layer of hydrophobic powder. Existing and emerging applications of liquid marbles require a contamination-free synthesis of liquid marbles with a high degree of reproducibility of their volume. Despite this requirement, the synthesis of liquid marbles has been still carried out manually. Manual production of liquid marbles leads to inconsistent volume and the possibility of contamination. The synthesis of liquid marbles with submicroliter volume is difficult to achieve and prone to large errors. This paper discusses the design and development of the first automated on-demand liquid marble generator with submicroliter capability. The device utilizes electrohydrodynamic pulling of liquid droplets on to a hydrophobic powder bed and subsequently coats them with the hydrophobic powder to synthesize liquid marbles of a desired volume.

Liquid marbles are liquid droplets covered with thin coating of hydrophobic powder. Aussillous and Quéré first reported the production of liquid marbles by coating water droplets with lycopodium grains covered with fluorinated silanes in 2001.1 Conventional liquid marbles are formed by rolling a small liquid droplet over a hydrophobic powder bed. However, liquid marbles covered by a monolayer of nanoparticles using the sol-gel technique have been reported.2–5 The hydrophobic powder coating on the liquid marble physically isolates the liquid droplet from the surrounding, providing a contamination-free environment to carry out sensitive chemical and biochemical reactions. Moreover, the coating helps the liquid droplet to keep its spherical shape even under relatively high mechanical stress.6 Mobility manipulation7–12 of liquid marbles is easier compared to liquid droplets without hydrophobic coating. Hence, liquid marbles find profound applications in biotechnology, biochemical reactions, cell biology, genetic research, bioengineering, drug discovery research, etc.9,13–17 Research studies have been advancing toward using liquid marbles as a digital microfluidic platform for DNA amplification techniques such as polymerase chain reaction (PCR) and loop-mediated isothermal amplification (LAMP).18,19 Existing and emerging applications of liquid marbles in various fields of digital microfluidics demand high-throughput generation of liquid marbles with precise control and consistency over their volume. Despite this requirement, the production of liquid marbles is currently still accomplished manually by depositing the liquid droplet on a hydrophobic powder bed using a micropipette and subsequent rolling. Limited throughput, inconsistency in volume, lack of repeatability, the possibility of human error, and contamination of liquid samples are some of the main disadvantages of this manual method. Moreover, the generation of liquid marbles with submicroliter volume is laborious and prone to significant volume inconsistencies. To the best of our knowledge, no other automated on-demand liquid marble generation system with submicroliter volume capacity has been reported in the literature.

In this paper, we report the design and development of an automated on-demand liquid marble generation system with submicroliter volume capacity using the concept of electrohydrodynamic pulling. The project utilizes electric field assisted deposition of liquid droplets on a hydrophobic powder bed and subsequent automated rolling for synthesizing liquid marbles of the desired volume. The proposed strategy is capable of synthesizing liquid marbles of submicroliter volume. In addition, this method avoids contamination of liquid samples due to manual interventions.

The paper is structured as follows. Section II explains the principle of operation of the proposed liquid marble generator and the design parameters of an on-demand liquid marble generator. Section III describes the fabrication of the automated on-demand liquid marble generator. Experimental evaluation of the developed on-demand liquid marble generator is explained in Sec. IV. Section V concludes the paper.

The fundamental idea behind the proposed liquid marble generator can be summarized as a three-step process: (i) generation of a liquid droplet using a mechanically actuated syringe; (ii) using an electric field to pull the liquid droplet of the desired volume formed at the tip of a syringe needle; and (iii) collecting the ejected droplet on a hydrophobic powder bed for subsequent automated rolling to synthesize liquid marbles. Knowledge of the physical parameters involved in these processes is essential to designing and optimizing the proposed liquid marble generator, and is explained as follows.

Figure 1(a) shows the schematic of a pendant droplet formed at the tip of a syringe needle with a high DC voltage applied between the needle and a metallic plate placed vertically below the needle. The force balance equation of a pendant drop at the tip of the syringe needle, when subjected to an electric field is given by20 

Fgravitation+FelectrostaticFsurfacetension=0,
(1)

where

Fgravitation=43πr3ρg,
(2)
Felectrostatic=12ε0SE2,
(3)

and

Fsurfacetension=2πRγ,
(4)

where r is the radius of the pendant droplet, ρ is the density of the liquid, g is the acceleration due to gravity, ε0 is the permittivity of the medium between the droplet surface formed at the tip of the needle and the cathode plate, S is the surface area of the droplet formed at the tip of the needle, E is the electric field strength between the droplet surface formed at the tip of the needle and the cathode plate, R is the outer radius of the syringe needle, and γ is the surface tension constant of the liquid. Equations (2)–(4) indicate that Fgravitation scales with the third order, Felectrostatic scales with the second order, and Fsurface tension scales with the first order of the droplet radius. This means that the electrostatic force is more favorable to counter the surface tension of smaller droplets. The pendant droplet would detach from the needle tip if the total downward force (Fgravitation + Felectrostatic) exceeds the surface tension force (Fsurface tension). It should be noted that the electric field strength E represented in Eq. (3) is a function of electrical properties (permittivity, polarizability, etc.) of the liquid droplet.21,22 However, Eqs. (1)–(4) indicate that for a particular pendant liquid droplet with radius r, Fgravitation and Fsurface tension are constant, whereas Felectrostatic is a function of the intensity electric field strength E between the droplet surface formed at the tip of the needle and the cathode plate. Alternatively, we may hypothesize that the intensity of the electric field at the tip of the needle is the parameter that determines the radius of the droplet detaching from the needle tip by overcoming the surface tension force. The radius of the droplet detaching from the needle decreases with increasing electric field strength.

FIG. 1.

Schematic of the pendant droplet under the electric field. (a) Without hydrophobic powder bed between anode and cathode and (b) with hydrophobic powder bed between anode and cathode.

FIG. 1.

Schematic of the pendant droplet under the electric field. (a) Without hydrophobic powder bed between anode and cathode and (b) with hydrophobic powder bed between anode and cathode.

Close modal

It has been previously reported in the literature that the electric field strength E at the tip of the needle is affected by the droplet formation. The electric field intensity at the tip of the needle with a liquid droplet can be estimated as20 

E=2VKrR+Rln4D2rKrR+R,
(5)

where V is the applied voltage, r is the radius of the droplet, R is the radius of the needle tip, D is the distance between the cathode plate and the tip of the needle, and K = 1.2 is a correction factor.

The droplet ejected from the tip of the syringe needle can be collected on a hydrophobic powder bed by placing the bed between the needle tip and the cathode plate [Fig. 1(b)]. However, the introduction of a hydrophobic powder bed between the syringe needle tip and cathode may alter the electric field at the tip of the needle and so the electrostatic force. The hydrophobic powder bed may act as a dielectric undergoing polarization in an electric field. The effective electric field at the tip of the needle is now reduced to

EEffective=EEPolarization,
(6)

where EPolarization is the electric field generated in the hydrophobic powder bed due to dielectric polarization. The reduced electric field would decrease the electrostatic force acting on the droplet

Felectrostatic=12ε0SEEPolarization2.
(7)

Hence, the volume of the droplet detached from the tip of the needle for a typical value of applied voltage is expected to be greater than that obtained in a setup without powder bed.

The numerical or analytical modeling of electric field intensity between anode and cathode with a hydrophobic powder bed between them is complex and out of the scope of this paper. However, the electric field strength EEffective between the tip of the needle and the aluminum plate can be considered as constant for a typical value of applied voltage as long as the mechanical arrangement of the setup and the ambient conditions remain unchanged. We can safely assume that the electric field EEffective at the tip of the needle and the electrostatic force Feffective acting on the droplet at the tip of the needle are only functions of the applied voltage “V.” Hence, the volume “v” of the droplet that can be pulled from the tip of the needle is also a function of the applied voltage. The relationship between the droplet volume ejected from the tip of the needle and the applied voltage can be determined experimentally. The details of the experiments are discussed as follows.

A primary experiment was carried out to find out the volume of droplets produced by various values of the applied voltage without the hydrophobic powder bed placed between the needle tip and the cathode plate. Figure 2(a) shows the schematic of the experimental setup. A vertically arranged 1-ml syringe with an inner diameter of 4.6 mm, attached with a blunt stainless-steel needle with an outer diameter of 250 µm, was used to produce liquid droplets of the desired volume. Deionized (DI) water was used as the working liquid. The syringe piston was pushed by a motorized micrometer stage (Zaber technologies) moving at a constant speed of 10.2 µm/s. The syringe needle was connected to the positive terminal of a high voltage source (HVPS2, 20 kV high voltage module, Eastern Voltage Research), while a thin aluminum plate (10 cm × 10 cm) was placed vertically below the needle and connected to the negative terminal of the voltage source. The distance between the tip of the needle and the aluminum plate was 10 cm. A high voltage, varying from 0 to 3.5 kV, was applied between the syringe needle tip and the aluminum plate, while the syringe piston was pushed at a speed of 10.2 µm/s. The formation and detachment of the droplet from the needle at various voltages are recorded using a high-speed camera (Photron FASTCAM SA3, attached with Nikon AF Micro Nikkor 60 mm f/2.8D lens) at a frame rate of 1000 fps. The liquid droplet ejection was observed to be in the microdripping regime from 0 to 3.4 kV and exhibited the formation of Taylor-cone beyond 3.4 kV. The microdripping regime is suitable for the design of a highly reproducible on-demand liquid droplet production. Images of the droplets just before their detachment from the needle tip were captured for evaluation. These images were processed in OpenDrop software23 to get the droplet volume before detaching from the needle tip.

FIG. 2.

Experimental setup of electrohydrodynamic pulling of the liquid droplet: (a) without PTFE powder bed between anode and cathode and (b) with PTFE powder bed between anode and cathode.

FIG. 2.

Experimental setup of electrohydrodynamic pulling of the liquid droplet: (a) without PTFE powder bed between anode and cathode and (b) with PTFE powder bed between anode and cathode.

Close modal

The experimental setup was subsequently rearranged by introducing a hydrophobic powder bed between the needle tip and the aluminum plate cathode. PTFE (polytetrafluoroethylene) powder (Sigma Aldrich) was used as the hydrophobic powder. A plastic rectangular spatula (35 mm × 35 mm × 8 mm) with negligible thickness, partially filled with PTFE powder, forms the PTFE powder bed. The PTFE powder bed was arranged at a distance of 2 cm vertically below the syringe needle to collect the ejected droplet. The horizontal high-speed camera was replaced by a compact CMOS camera (ximea xiQ), attached with 0.5× telecentric lens (Edmund Optics, EO-63074) for hardware simplicity. A new vertical CMOS camera (Edmund Optics, E-05012C) attached with 1× telecentric lens (Edmund Optics, EO-58430) was introduced in the setup to record the images of the liquid droplet deposited on the PTFE bed and the subsequently synthesized liquid marble. Figure 2(b) shows the experimental setup. High voltages varying from 0 to 3.4 kV were applied in steps, and the droplets were collected on the PTFE powder bed. These droplets were manually rolled on the PTFE powder bed to coat the droplet with PTFE powder and form liquid marbles. The liquid droplet and the PTFE powder were kept under an ionizing air blower (STATIC-ST101A) for a few seconds before starting the rolling process. This step ensures the charge neutralization of the PTFE powder and the droplet. Charge neutralization of the PTFE powder and droplet allows for the error-free rolling process and uniform coating of the powder over the liquid droplet. Images of the deposited droplet and the corresponding liquid marble were captured by the vertical camera. These images were subsequently processed using ImageJ software to evaluate their respective volume. The variation of the volume of the droplets with respect to the applied voltage in the first and second experiments is shown in Fig. 3(a). The droplet volume obtained in the first experiment (without PTFE powder bed between the anode and cathode) is smaller than the droplet volume observed in the second experiment for the applied voltage up to a threshold value, after which the volumes of both cases are approximately equal. The effect of reduced electric field intensity was observed to be significant at lower voltage resulting in relatively larger differences in droplet volumes between the first and the second experiment. The difference in volume is smaller at a higher voltage. Curve fitting was performed with the data obtained from the experiments using the MATLAB curve fitting tool. A generalized relationship between the volume of the droplet deposited on the PTFE powder bed and the applied voltage was obtained as

v=3.67×sin(0.19×V+1.90),
(8)

where v is the volume of the droplet in microliter and V is the applied voltage in kilovolt.

FIG. 3.

(a) Variation of droplet volumes with applied voltage. (b) Graph showing the calculated and observed values of time of detachment of the liquid droplet for various values of droplet volume.

FIG. 3.

(a) Variation of droplet volumes with applied voltage. (b) Graph showing the calculated and observed values of time of detachment of the liquid droplet for various values of droplet volume.

Close modal

The volume of the droplet at the tip of the needle is a function of time, since the droplet formation is caused by the continuous uniform actuation of the syringe piston. The rate of change of volume at the tip of the needle can be given as

dv(t)dt=πrsyringe2dhdt,
(9)

where rsyringe is the inner diameter of the syringe cavity and dhdt is the downward velocity of the syringe piston. The volume of the droplet on the needle tip at any instant of time t can be obtained by solving Eq. (9), substituting the constant values of rsyringe = 2.3 mm and dhdt=10.23μm/s. The volume of the droplet v(t) at any instance t is given as

vt=0.17×t.
(10)

Equation (10) can be used to find out the time “t” at which a particular volume of the droplet detaches from the needle tip. The calculated and observed values of t for various volumes of liquid droplets deposited on the PTFE powder bed are shown in Fig. 3(b).

A fully automated on-demand liquid marble generation system was designed and tested by considering the design parameters discussed in Sec. II A. The automated liquid marble generator consists of two main subsystems, namely, (i) the droplet generation and deposition subsystem and (ii) the hydrophobic powder coating subsystem. The details of the fabrication process of each of these subsystems are explained as follows.

A vertically arranged 1-ml syringe with an inner diameter of 4.6 mm, attached with a metallic needle of outer diameter 250 µm, was used to produce liquid droplets of a desired volume. The motorized micrometer stage moving at a velocity of 10.2 µm/s actuated the syringe piston. The micrometer stage was controlled by an Arduino UNO microcontroller board through a relay. The liquid droplet started forming at the tip of the syringe needle upon the actuation of the micrometer stage by the microcontroller. The volume of the droplet at the tip of the syringe needle v(t) follows Eq. (10).

The deposition of the droplet generated at the tip of the needle on the PTFE powder bed is done by applying a suitable electric field between the syringe needle and an aluminum plate placed 10 cm vertically below the needle tip. The Arduino UNO microcontroller is programmed to receive the volume input through a computer interface. The microcontroller calculates the high voltage required to pull the specific volume of a liquid droplet using the relationship shown in Eq. (8). The microcontroller subsequently uses the calibration data of the high voltage module to determine the DC input voltage required for the high voltage module. A digital equivalent of the calculated DC input voltage is generated by the microcontroller and converted into an analog DC voltage using the MCP 4725 digital to analog converter. The analog DC input voltage is applied to the high voltage DC module through a buffer amplifier designed using the LM 358 operational amplifier and TIP 120 Darlington power transistor to generate the high voltage required to electrohydrodynamically pull the required volume of the droplet. The buffer amplifier was biased at 12 V, 1 A DC using a programmable power supply (KEITHLY, 30 V, 5 A). The accuracy of the high voltage applied between the tip of the needle and aluminum plate determines the accuracy of the droplet volume deposited on the PTFE powder bed. Hence, feedback is provided from the high voltage module to the microcontroller using a 1000:1 voltage probe (40 kV HV probe, TENMA 72-3040) and a proportional-integral-derivative (PID) control action with 5% tolerance which was performed by the microcontroller to ensure the accuracy of the generated high voltage. It should be noted that the microcontroller actuates the syringe piston only after setting the suitable high voltage between the anode and cathode. The droplet detaches from the needle tip when its volume v is equal to the value, which can be pulled by the applied voltage according to Eq. (8). The pulled liquid droplet would be deposited on to a PTFE powder bed placed 2 cm vertically below the syringe needle tip.

The droplet deposited on the PTFE powder bed has to be coated with the PTFE powder to form a liquid marble. This was achieved by oscillating the PTFE powder bed 20 times between 0° and 40° with an angular velocity of 1.2 rad/s. The optimum value of the angular velocity was obtained by the trial-and-error method. A servomotor (Tower Pro-MG 995) controlled by the Arduino UNO microcontroller was used to oscillate the PTFE powder bed. It should be noted that the PTFE powder bed should start oscillating only after (i) the droplet is deposited on the powder bed and (ii) the powder bed and droplet being neutralized by the ionizing air blower. The time t required to generate a specific volume of the droplet at the tip of the needle can be calculated using Eq. (10). An upper tolerance of 10% was allotted to the oscillation subsystem to address the errors due to the mechanical inertia of the motorized micrometer stage and the syringe piston. The ionizing air blower was turned on subsequently by the microcontroller for 8 s through a relay. The servo motor starts the oscillation process after the air blower is stopped. Figure 4 shows the schematic of the fully automated on-demand liquid marble generator.

FIG. 4.

Schematic of the experimental setup of on-demand liquid marble generator. Photograph of the setup is shown in the inset.

FIG. 4.

Schematic of the experimental setup of on-demand liquid marble generator. Photograph of the setup is shown in the inset.

Close modal

The on-demand liquid marble generator in Sec. III was employed to generate liquid marbles of various volumes. Liquid marbles of volumes ranging from 700 nl to 5 µl were generated automatically with an average error of 8.6% in volume. Figure 5 shows the liquid droplet volumes generated by the automated on-demand liquid marble generator against the expected volumes. The error obtained in the droplet volume compared to the expected value of droplet volume can be attributed to the errors of the high voltage module. This error can be reduced by bringing down the tolerance band of the PID control algorithm from 5% to a suitable value. Reduction in the tolerance band of the PID algorithm would generate the high voltage values precisely, which in turn increases the accuracy in droplet volumes. However, narrowing the PID tolerance band may increase the settling time of the PID controller so that the time required to generate the required precise voltage values would be increased. This may limit the throughput of the proposed liquid marble generator.

FIG. 5.

Graph showing the expected and obtained volume of droplets using the automated on-demand liquid marble generator. The micrographs of droplets and corresponding liquid marbles are also depicted.

FIG. 5.

Graph showing the expected and obtained volume of droplets using the automated on-demand liquid marble generator. The micrographs of droplets and corresponding liquid marbles are also depicted.

Close modal

This paper reported the design and development of an on-demand liquid marble generator with submicroliter volume capacity. The liquid marble generator utilized the electric field to get control over the volume of liquid marbles. The volume of the liquid droplet was found to be a function of the applied electric field. The relationship between the droplet volume and the applied voltage was obtained experimentally. A simple theoretical model to estimate the time required to generate a specific volume of the droplet was also developed. These parameters were utilized to implement the on-demand liquid marble generator. Mechanical and electrical subsystems were designed and fabricated to automate the process of droplet deposition and coating. The on-demand liquid marble generator was successfully tested, and liquid marbles of a desired volume were synthesized with an average error of 8.6%.

Future studies are extended to be the development of a generalized theoretical model representing the relationship between the droplet volume and the applied voltage, with a hydrophobic powder bed between anode and cathode. The physical, chemical, and electrical properties of the hydrophobic powder are expected to play a significant role in the high throughput production, the quality of coating, and also the accuracy of the droplet volume. This theoretical model can be used in the already developed on-demand liquid marble generator to produce liquid marbles with nano/picoliter volume capacities and better operating volume ranges. The optimization of the PID control algorithm can also be carried out to achieve better accuracy on the volume of the droplet without sacrificing the throughput capability. Though the fundamental electrohydrodynamic pulling strategy of liquid droplet offers a high throughput production of liquid marbles, the proposed single hydrophobic bed beneath the syringe needle limits the throughput capability to a certain extent. A multiple hydrophobic powder bed along with suitable marble transfer mechanism would result in increasing the throughput capability of the proposed instrument. Any of the existing liquid marble mobility manipulation techniques can be suitably engineered to use as a marble transfer mechanism.

The authors acknowledge the Australian Research Council for funding support through Grant No. DP170100277. This work was performed in part at the Queensland node of the Australian National Fabrication Facility, a company established under the National Collaborative Research Infrastructure Strategy to provide nano- and microfabrication facilities for Australia’s researchers.

1.
P.
Aussillous
and
D.
Quéré
,
Nature
411
,
924
(
2001
).
2.
X.
Li
,
Y.
Wang
,
J.
Huang
,
Y.
Yang
,
R.
Wang
,
X.
Geng
, and
D.
Zang
,
Appl. Phys. Lett.
111
,
261604
(
2017
).
3.
X.
Li
,
R.
Wang
,
H.
Shi
, and
B.
Song
,
Appl. Phys. Lett.
113
,
101602
(
2018
).
4.
X.
Li
,
R.
Wang
,
S.
Huang
,
Y.
Wang
, and
H.
Shi
,
Soft Matter
14
,
9877
(
2018
).
5.
K.
Lin
,
R.
Chen
,
L.
Zhang
,
D.
Zang
,
X.
Geng
, and
W.
Shen
,
ACS Appl. Mater. Interfaces
11
,
8789
(
2019
).
6.
H.-N.
Polwaththe-Gallage
,
C. H.
Ooi
,
J.
Jin
,
E.
Sauret
,
N.-T.
Nguyen
,
Z.
Li
, and
Y.
Gu
,
Appl. Phys. Lett.
114
,
043701
(
2019
).
7.
L.
Mahadevan
and
Y.
Pomeau
,
Phys. Fluids
11
,
2449
(
1999
).
8.
E.
Bormashenko
,
R.
Pogreb
,
R.
Balter
,
O.
Gendelman
, and
D.
Aurbach
,
Appl. Phys. Lett.
100
,
151601
(
2012
).
9.
Y.
Xue
,
H.
Wang
,
Y.
Zhao
,
L.
Dai
,
L.
Feng
,
X.
Wang
, and
T.
Lin
,
Adv. Mater.
22
,
4814
(
2010
).
10.
M. K.
Khaw
,
C. H.
Ooi
,
F.
Mohd-Yasin
,
R.
Vadivelu
,
J. S.
John
, and
N. T.
Nguyen
,
Lab Chip
16
,
2211
(
2016
).
11.
L.
Zhang
,
D.
Cha
, and
P.
Wang
,
Adv. Mater.
24
,
4756
(
2012
).
12.
Y.
Zhao
,
Z.
Xu
,
M.
Parhizkar
,
J.
Fang
,
X.
Wang
, and
T.
Lin
,
Microfluid. Nanofluid.
13
,
555
(
2012
).
13.
K. A.
Heyries
,
C.
Tropini
,
M.
Vaninsberghe
,
C.
Doolin
,
O. I.
Petriv
,
A.
Singhal
,
K.
Leung
,
C. B.
Hughesman
, and
C. L.
Hansen
,
Nat. Methods
8
,
649
(
2011
).
14.
H.
Zhang
,
G.
Jenkins
,
Y.
Zou
,
Z.
Zhu
, and
C. J.
Yang
,
Anal. Chem.
84
,
3599
(
2012
).
15.
T.
Arbatan
,
L.
Li
,
J.
Tian
, and
W.
Shen
,
Adv. Healthcare Mater.
1
,
80
(
2012
).
16.
F.
Sarvi
,
T.
Arbatan
,
P. P. Y.
Chan
, and
W.
Shen
,
RSC Adv.
3
,
14501
(
2013
).
17.
M. C.
Serrano
,
S.
Nardecchia
,
M. C.
Gutierrez
,
M. L.
Ferrer
, and
F.
del Monte
,
ACS Appl. Mater. Interfaces
7
,
3854
(
2015
).
18.
K. R.
Sreejith
,
C. H.
Ooi
,
D. V.
Dao
, and
N.-T.
Nguyen
,
RSC Adv.
8
,
15436
(
2018
).
19.
K. R.
Sreejith
,
C. H.
Ooi
,
J.
Jin
,
D. V.
Dao
, and
N.-T.
Nguyen
,
Lab Chip
18
,
3717
(
2018
).
20.
Y.-J.
Kim
,
H.
Seo Ko
,
S.
Lee
,
S.
Uk Son
,
D.
Jung
, and
D.
Byun
,
J. Korean Phys. Soc.
51
,
42
(
2007
).
21.
E.
Bormashenko
,
R.
Pogreb
,
T.
Stein
,
G.
Whyman
,
M.
Schiffer
, and
D.
Aurbach
,
J. Adhes. Sci. Technol.
25
,
1371
(
2012
).
22.
P. K.
Notz
and
O. A.
Basaran
,
J. Colloid Interface Sci.
213
,
218
(
1999
).
23.
J. D.
Berry
,
M. J.
Neeson
,
R. R.
Dagastine
,
D. Y. C.
Chan
, and
R. F.
Tabor
,
J. Colloid Interface Sci.
454
,
226
(
2015
).