Carbon nanotube (CNT) fibers are used in various applications, such as electrical cables, supercapacitors, physical sensors, artificial muscles, and electronic devices, due to their excellent mechanical, thermal, and electrical properties. Herein, the orientation-dependent electrical characteristics of a CNT fiber array were investigated. A force resistive sensor with good sensitivity and reliable operation was developed using the array and integrated with a real-time data storage and wireless monitoring system. In addition, a CNT fiber-based position sensor was developed for application in smart fabrics. This study introduces an easy-to-manufacture, low-cost, robust CNT fiber-based sensing platform that can be used with an open-source microcontroller for integration with the Internet of things.

The Internet of things (IoT), which is one of the major drivers of the Fourth Industrial Revolution, requires development of advanced sensors and sensing platforms capable of detecting, monitoring, and predicting physical signals in real time from external environments and human bodies.1–5 High-performance sensors and cost-effective and reliable IoT systems can be achieved using robust and low-cost nanomaterials for the realization of smart factories, smart cities, autonomous vehicles, healthcare, and security systems.1,2,6 Carbon nanotubes (CNTs) exhibit exceptional electrical, thermal, and mechanical properties, with excellent corrosion resistance and potential for mass production.7 An individual CNT has a tensile strength of about 100 GPa, a current density of up to 109 A/cm2, and a thermal conductivity of 3500 W m−1 K−1 at room temperature. Therefore, CNTs have been implemented in many applications, including advanced composite materials; transistors for microelectronics, energy storage, and biotechnology; and environmental applications. Traditional fibers, such as those of silk, cotton, and flax, have widely been used over the course of human history because they can be easily knitted, weaved, knotted, and bound.8 Recently, smart fibers, functionalized by various treatments or combined with various functional materials, have been developed for sensing, actuating, displaying, and computing applications.8,9 In particular, a CNT-fiber, a macroscopic assembly of CNTs, is currently in the spotlight due to its superior mechanical strength with very good structural flexibility and good thermal and electrical properties inherited from those of individual CNTs.9 Carbon nanotube fibers can be synthesized by forest, direct, and solution spinning, and they can be transformed into electrical cables, supercapacitors, physical sensors, artificial muscles, and solar cells.10,11

Herein, we report the orientation-dependent electrical properties of the CNT fiber arrays fabricated by the direct spinning method. These arrays were used to develop high-performance force resistive sensors (FRSs) and knittable position sensors that are appropriate for IoT systems and smart fabrics. Finally, the fabricated sensors were combined with real-time data storage and a monitoring system using an Arduino open-source microcontroller.

CNT-fibers were prepared by the direct spinning method, in which CNTs are synthesized and spun simultaneously in the gas phase using a high temperature reactor, as shown in Fig. 1(a).13 The direct spinning method is more advantageous for mass production than other methods because it is a one-step and continuous process. In addition, highly efficient production of CNT-fibers with still attractive properties can be achieved by synthesizing CNT-fibers with a high linear density.12 A CNT fiber consists of a tremendous number of individual CNTs; the Raman spectrum of an assembly of CNT fibers is presented in Fig. 1(b). Two intense peaks with a good IG/ID ratio of about 3.6 are typical of the Raman spectra of CNTs.14,15 Disorder, impurities, and imperfections of graphitic domains relate to the first peak (D band) at 1330 cm−1, while the second peak (G band) at 1560 cm−1 is associated with the E2g vibrational mode of the unit cell of hexagonal crystalline graphene.

FIG. 1.

(a) Scanning electron microscope images of carbon nanotube (CNT) fibers synthesized by the direct spinning method. The CNT fiber consisted of a tremendous number of carbon nanotubes. (b) Raman spectrum of the CNT fibers.

FIG. 1.

(a) Scanning electron microscope images of carbon nanotube (CNT) fibers synthesized by the direct spinning method. The CNT fiber consisted of a tremendous number of carbon nanotubes. (b) Raman spectrum of the CNT fibers.

Close modal

The CNT fibers were placed on a glass substrate in two orientations, i.e., parallel to and in series with the electrode orientation [Figs. 2(a) and 2(b)]. The electrical properties of the fabricated devices based on the differently aligned CNT fiber arrays were measured as shown in Fig. 2(c). Notably, the sheet resistance of the CNT fiber array aligned in series (Rseries = V/I × W/L, where W and L are the width and length of the CNT fiber array, about 0.3 Ω sq−1) was about 100 times smaller than that of the array aligned in parallel (Rparallel, about 31 Ω sq−1). Figure 3 shows the device structures with the expected electrical parameters and equivalent circuit models. In the figure, RCntF-L and RCntF-W denote the sheet resistance of the CNT fiber in the longitudinal and width directions, respectively. Rseries is composed of several RCntF-L connected in parallel and is calculated as 1/(∑1/RCntF-L). In contrast, the device shown in Fig. 3(a) consists of RCntF-W with interfacial resistance between the CNT fibers (Rinterface). Rparallel is given by ∑RCntF-W + ∑Rinterface. Rseries was much smaller than Rparallel, possibly because Rinterface could be much higher than RCntF-W and RCntF-L. The sheet resistance of CNT fiber arrays can also be optimized by controlling the densification of individual CNT fibers.16–18 

FIG. 2.

The carbon nanotube (CNT) fiber array placed on a glass substrate with two distinct orientations, i.e., (a) in parallel and (b) in series relative to the electrodes. (c) Voltage–current characteristics of the CNT fiber array with different orientations.

FIG. 2.

The carbon nanotube (CNT) fiber array placed on a glass substrate with two distinct orientations, i.e., (a) in parallel and (b) in series relative to the electrodes. (c) Voltage–current characteristics of the CNT fiber array with different orientations.

Close modal
FIG. 3.

(a) Illustrations of the carbon nanotube fiber-based array device structures aligned (a) in parallel and (b) in series with their equivalent circuit models.

FIG. 3.

(a) Illustrations of the carbon nanotube fiber-based array device structures aligned (a) in parallel and (b) in series with their equivalent circuit models.

Close modal

Figures 4 and 5 show the performance of the FRS fabricated using a CNT fiber array as an active layer. An active sensing array having a parallel orientation was placed on a glass substrate. Then, CNT fibers aligned in series on a flexible polyethylene terephthalate (PET) substrate were placed on the active layer with a supporting spacer [Fig. 4(a)]. When force was applied to the flexible PET substrate, the CNT fibers aligned in series (of much smaller resistance) touched the active layer (of much higher resistance), which significantly changed the electrical properties measured through the active sensing element.19  Figure 4(b) shows the operating mechanism of the fabricated force sensor. Under applied pressure, a new conducting path was created via the CNT fibers oriented in series on the PET substrate, which effectively reduced the resistance of the active layer. In addition, pressure induced some lateral deformations in individual CNT fibers and increased the effective contact area between CNT fibers aligned in parallel, which reduced Rinterface.

FIG. 4.

(a) Fabricated force resistive sensor based on the carbon nanotube fiber-based array. (b) Illustration of the operating mechanism of the fabricated force sensor.

FIG. 4.

(a) Fabricated force resistive sensor based on the carbon nanotube fiber-based array. (b) Illustration of the operating mechanism of the fabricated force sensor.

Close modal
FIG. 5.

(a) Voltage–current characteristics of the force resistive sensor as a function of pressure. (b) Change of resistance in the force sensor as a function of applied pressure. (c) Response and recovery characteristics of the force sensor during two levels of cyclic pressure loading/unloading.

FIG. 5.

(a) Voltage–current characteristics of the force resistive sensor as a function of pressure. (b) Change of resistance in the force sensor as a function of applied pressure. (c) Response and recovery characteristics of the force sensor during two levels of cyclic pressure loading/unloading.

Close modal
In the case of the device structures placing CNT fibers in series for both bottom and top sides, there was no substantial variation in electrical current, since the resistance of the CNT fiber in series was already very low. The electrical current measured through the force sensor dramatically increased with increasing pressure [Fig. 5(a)]. The sensitivity (S) of force sensors can be determined as follows:20,
S=δ[(R0Rp)/R0]δP=δ(ΔR/R0)δP,
(1)
where P, R0, and Rp are the applied force, resistance at base pressure, and resistance under pressure, respectively. The sensitivity of the CNT fiber-based force sensor was about 0.6 kPa−1 in the low-pressure regime (<1 kPa) [Fig. 5(b)], comparable with that of CNT-based tactile sensors and graphene-based force sensors.20,21 However, the S value was reduced to about 0.025 kPa−1 at higher pressures. Sensitivity could be enhanced by controlling the CNT fiber diameter and structural design of the aligned CNT fibers. Post-treatments, for example, to eliminate impurities, densify, and manipulate fiber alignment could be used to tune the electrical properties of CNT fibers, and they may also improve the S value and sensing range. The response and recovery characteristics of the force sensor were measured under two levels of cyclic pressure loading/unloading. Figure 5(c) shows the good reproducibility and repeatability of the sensing performance.

The technical issues associated with storage, real-time analysis, and unlimited communication of big data from sensor networks in various environments can be overcome with cloud computing-based IoT systems via a wireless network protocol, such as Wi-Fi.3–5 A simple system consisting of a CNT fiber-based FRS, an Arduino board with a Wi-Fi module (ESP8266 NodeMCU), and an Arduino IoT cloud is shown in Fig. 6(a), and its operation is presented in Fig. 6(b). The sensor was connected to a commercial 30-Ω resistor operating at 3.3 V in series, which acted as a voltage divider. The resistance of the sensor decreased with increasing applied pressure, and the output voltage (Vout) changed according to Vout = 3.3 V × 30 Ω/(resistance of the FRS + 30 Ω). The Arduino microcontroller detected Vout via the analog input (A0) port and simultaneously sent the measured data to the Arduino cloud server via Wi-Fi. The measured data could also be monitored in real time by a smartphone. Figure 6(b) shows the sensing data obtained during repeated pressure and release cycles; the data were automatically stored in the Arduino IoT cloud and a smartphone. The electrical connections and programming codes for building the Arduino system are presented in the supplementary material, Fig. S1.

FIG. 6.

(a) Simple sensing system enabling real-time data monitoring and storage in the cloud consisting of a carbon nanotube fiber-based force resistive sensor, an Arduino board with a Wi-Fi module, and the Arduino IoT cloud. (b) Demonstration showing the simultaneous data storage in the Arduino IoT cloud and a smartphone.

FIG. 6.

(a) Simple sensing system enabling real-time data monitoring and storage in the cloud consisting of a carbon nanotube fiber-based force resistive sensor, an Arduino board with a Wi-Fi module, and the Arduino IoT cloud. (b) Demonstration showing the simultaneous data storage in the Arduino IoT cloud and a smartphone.

Close modal

Figure 7 shows the proof-of-concept CNT fiber-based potentiometric position sensor fabricated using the Arduino Uno microcontroller. Position sensors can detect the movement or location of an object and transform this into an electrical signal for monitoring and control.22 The simply fabricated, fabric-integrated position sensor and its equivalent circuit are shown in Fig. 7(a). The Vin (=3.3 V) and ground (=0 V; GND) were connected at both ends of the horizontal CNT fiber, which was sewn and fixed to the fabric. The CNT fiber sewn in the vertical direction served as an electrode to send a voltage signal to the Arduino board and was connected to the GND through the load resistance (Rload = 10 kΩ) to transmit a signal in a stable state in the absence of pressure. When the pressure was applied to a vertically oriented CNT fiber, it contacted a horizontal CNT fiber and a voltage drop was induced according to the contact position (Vout = 3.3 V × Rcrt/Rtot, where Rtot and Rcrt are the total resistance of a horizontal CNT fiber and the resistance of a CNT fiber at the point of contact, respectively), thereby determining the contact position. The interface between CNT fibers was coated with silver conductive paste to ensure stable and reliable contact. The detailed electrical connections to the Arduino board and a liquid-crystal display (LCD) are shown in Fig. 7(b), and the successful operation of the CNT fiber-based position sensor is shown in Fig. S2.

FIG. 7.

(a) Proof-of-concept carbon nanofiber-based potentiometric position sensor with its equivalent circuit model. (b) Detailed electrical connections to the Arduino board and a liquid-crystal display for operating the position sensor.

FIG. 7.

(a) Proof-of-concept carbon nanofiber-based potentiometric position sensor with its equivalent circuit model. (b) Detailed electrical connections to the Arduino board and a liquid-crystal display for operating the position sensor.

Close modal

The aligned orientation-dependent resistance properties of CNT fibers, synthesized by the direct spinning method, were investigated. The sheet resistance of a CNT fiber array aligned in series (Rseries) was about 100-times smaller than that of an array aligned in parallel (Rparallel), possibly because Rinterface could be much higher than the sheet resistance of the CNT fiber in the longitudinal and width directions. An FRS with the CNT fiber array was also fabricated, and its performance was demonstrated; it displayed a good sensitivity of about 0.6 kPa−1 and reliable response/recovery characteristics. In addition, the force sensor was combined with real-time data storage and a monitoring system for the development of IoT cloud systems. Furthermore, a knittable CNT fiber-based potentiometric position sensor suitable for smart fabrics was implemented and operated using an Arduino microcontroller and LCD monitor. There is a possibility that multi-functional force and position sensors can be also implemented with moisture-dependent electrical characteristics of CNT fibers.23,24

Direct-spun CNT fibers with a high linear density of about 12 tex (g/km) were provided by JEIO Co., Ltd. (South Korea).12 Raman spectra were obtained using a Horiba LabRAM micro-Raman instrument with a laser wavelength of 514 nm. The electrical performance of FRSs based on CNT fibers was measured using a Keithley 2400 instrument with a LABVIEW software interface for data acquisition and analysis. The Arduino IoT cloud system (https://cloud.arduino.cc/) and an ESP8266 NodeMCU microcontroller were used to operate the force sensors, which were equipped with Wi-Fi communication. Position sensors made of CNT fibers were also demonstrated using an Arduino Uno board. Detailed programming codes for operating the Arduino board and using the IoT cloud system are described in the supplementary material, Figs. S1–S3.

See the supplementary material for details with respect to programming codes for an Arduino board and a wireless network describing the Arduino IoT cloud system.

This work was supported by the National Research Foundation of Korea (NRF) (Grant No. 2017M3A7B4049167), the Korean Institute of Science and Technology (KIST) Open Research Program (2E32631), and the Technology Innovation Program (20017548) funded by the Ministry of Trade, Industry & Energy (MOTIE, Korea). The authors acknowledge JEIO Co., Ltd. for providing CNTFs for this study.

The authors have no conflicts to disclose.

Dae-Young Jeon: Conceptualization (lead); Formal analysis (lead); Supervision (lead); Writing – original draft (lead); Writing – review & editing (lead). Jimin Park: Methodology (equal); Resources (equal). Changwoo Lee: Methodology (equal); Resources (equal). Seung Min Kim: Funding acquisition (equal); Methodology (equal).

The data that support the findings of this study are available from the corresponding authors upon reasonable request.

1.
J. C.
Yeo
and
C. T.
Lim
,
Microsyst. Nanoeng.
2
(
1
),
16043
(
2016
).
2.
J. A.
Cardenas
,
J. B.
Andrews
,
S. G.
Noyce
, and
A. D.
Franklin
,
Nano Futures
4
(
1
),
012001
(
2020
).
3.
I. U.
Din
,
M.
Guizani
,
S.
Hassan
,
B.-S.
Kim
,
M. K.
Khan
,
M.
Atiquzzaman
, and
S. H.
Ahmed
,
IEEE Access
7
,
7606
7640
(
2018
).
4.
I. C. L.
Ng
and
S. Y. L.
Wakenshaw
,
Int. J. Res. Mark.
34
(
1
),
3
21
(
2017
).
5.
D.
Gil
,
A.
Ferrández
,
H.
Mora-Mora
, and
J.
Peral
,
Sensors
16
(
7
),
1069
(
2016
).
6.
H.
Souri
,
H.
Banerjee
,
A.
Jusufi
,
N.
Radacsi
,
A. A.
Stokes
,
I.
Park
,
M.
Sitti
, and
M.
Amjadi
,
Adv. Intell. Syst.
2
(
8
),
2000039
(
2020
).
7.
M. F. L.
De Volder
,
S. H.
Tawfick
,
R. H.
Baughman
, and
A. J.
Hart
,
Science
339
(
6119
),
535
539
(
2013
).
8.
H.
Wang
,
Y.
Zhang
,
X.
Liang
, and
Y.
Zhang
,
ACS Nano
15
(
8
),
12497
12508
(
2021
).
9.
J.
Di
,
X.
Zhang
,
Z.
Yong
,
Y.
Zhang
,
D.
Li
,
R.
Li
, and
Q.
Li
,
Adv. Mater.
28
(
47
),
10529
10538
(
2016
).
10.
S.-H.
Lee
,
J. H.
Park
, and
S. M.
Kim
,
J. Korean Ceram. Soc.
58
(
2
),
148
159
(
2021
).
11.
A.
Lekawa‐Raus
,
J.
Patmore
,
L.
Kurzepa
,
J.
Bulmer
, and
K.
Koziol
,
Adv.Funct. Mater.
24
(
24
),
3661
3682
(
2014
).
12.
D.-M.
Lee
,
J.
Park
,
J.
Lee
,
S.-H.
Lee
,
S.-H.
Kim
,
S. M.
Kim
, and
H. S.
Jeong
,
Carbon
172
,
733
741
(
2021
).
13.
Y.-L.
Li
,
I. A.
Kinloch
, and
A. H.
Windle
,
Science
304
(
5668
),
276
278
(
2004
).
14.
H.
Murphy
,
P.
Papakonstantinou
and
T. I. T.
Okpalugo
,
J. Vac. Sci. Technol. B
24
(
2
),
715
720
(
2006
).
15.
V.
Duc Chinh
,
G.
Speranza
,
C.
Migliaresi
,
N.
Van Chuc
,
V.
Minh Tan
, and
N.-T.
Phuong
,
Sci. Rep.
9
(
1
),
5667
(
2019
).
16.
Z.
Wang
,
T. J.
Mun
,
F. M.
Machado
,
J. H.
Moon
,
S.
Fang
,
A. E.
Aliev
,
M.
Zhang
,
W.
Cai
,
J.
Mu
, and
J. S.
Hyeon
,
Adv. Mater.
34
(
27
),
2201826
(
2022
).
17.
S.
Oh
,
K. J.
Kim
,
B.
Goh
,
C. L.
Park
,
G. D.
Lee
,
S.
Shin
,
S.
Lim
,
E. S.
Kim
,
K. R.
Yoon
, and
C.
Choi
,
Adv. Sci.
9
(
32
),
2203767
(
2022
).
18.
M.
Zhang
,
W.
Cai
,
Z.
Wang
,
S.
Fang
,
R.
Zhang
,
H.
Lu
,
A. E.
Aliev
,
A. A.
Zakhidov
,
C.
Huynh
, and
E.
Gao
,
Nat. Energy
8
,
203
213
(
2023
).
19.
D.-Y.
Jeon
,
S. J.
Park
,
T. Y.
Lee
, and
G.-T.
Kim
,
Sens. Actuators, A
332
,
113209
(
2021
).
20.
H.
Tian
,
Y.
Shu
,
X.-F.
Wang
,
M. A.
Mohammad
,
Z.
Bie
,
Q.-Y.
Xie
,
C.
Li
,
W.-T.
Mi
,
Y.
Yang
, and
T.-L.
Ren
,
Sci. Rep.
5
(
1
),
8603
(
2015
).
21.
S. Y.
Kim
,
S.
Park
,
H. W.
Park
,
D. H.
Park
,
Y.
Jeong
, and
D. H.
Kim
,
Adv. Mater.
27
(
28
),
4178
4185
(
2015
).
22.
D. S.
Nyce
,
Position Sensors
(
John Wiley & Sons
,
2016
).
23.
L.
Liu
,
X.
Ye
,
K.
Wu
,
R.
Han
,
Z.
Zhou
, and
T.
Cui
,
Sensors
9
(
3
),
1714
1721
(
2009
).
24.
H.-S.
Kim
,
J.-H.
Kang
,
J.-Y.
Hwang
, and
U. S.
Shin
,
Nano Convergence
9
(
1
),
35
(
2022
).

Supplementary Material