lable at ScienceDirect

Materials Today Physics 13 (2020) 100199

Contents lists avai

Materials Today Physicsjournal homepage: https: / /www.journals .e lsevier .com/

mater ia ls- today-physics

Epidermal electronics for respiration monitoring via thermo-sensitivemeasuring

Y. Liu a, L. Zhao a, R. Avila b, C. Yiu a, T. Wong a, Y. Chan a, K. Yao a, D. Li a, Y. Zhang a, **,W. Li c, ***, Z. Xie d, ****, X. Yu a, *

a Department of Biomedical Engineering, City University of Hong Kong, Hong Kong, 999077, People's Republic of Chinab Department of Mechanical Engineering, Northwestern University, Evanston, IL, 60208, USAc Department of Mechanical Engineering, City University of Hong Kong, Hong Kong, 999077, People's Republic of Chinad State Key Laboratory of Structural Analysis for Industrial Equipment, Department of Engineering Mechanics, International Research Center forComputational Mechanics, Dalian University of Technology, Dalian, 116024, People's Republic of China

a r t i c l e i n f o

Article history:Received 9 January 2020Received in revised form14 February 2020Accepted 22 February 2020Available online 29 February 2020

Keywords:Wearable electronicsThermo-sensitivityRespiration sensorStretchable electronics

* Corresponding author.** Corresponding author.*** Corresponding author.**** Corresponding author.

E-mail addresses: yt.zhang@cityu.edu.hk (Y. Zhangzxie@dlut.edu.cn (Z. Xie), xingeyu@cityu.edu.hk (X. Y

https://doi.org/10.1016/j.mtphys.2020.1001992542-5293/© 2020 Elsevier Ltd. All rights reserved.

a b s t r a c t

The depth and rate of human respiration reveal important and diverse sets of physiological informationfor evaluating human health. Here, we introduce an ultrathin, skin-integrated respiration sensor basedon the thermal convection effect. The device features a filamentary fractal design of the gold heatingelectrode, a mini sensor (0.6 mm � 0.3 mm � 0.23 mm) with high thermal sensitivity and an ultrasoftencapsulation package to enhance the overall flexibility and biaxial stretchability of the system.Adjusting the input power of the heating electrode, i.e., increasing the temperature difference betweenthe thermal sensor and environment, can further improve the sensitivity of the respiration sensor. Thereal-time monitoring respiration sensor can competently distinguish various breathing patterns (sitting,frightening, sleeping, meditating, and gasping) through breath rate/depth of detection subjects. Inaddition, the respiration sensor can effectively capture, in real time, the respiration of a volunteer whileexercising, resting, or sleeping for prolonged periods of time. The combination of advanced mechanics,high sensitivity, and good stability make this respiration sensor a great candidate for potential use inreal-time monitoring of human health.

© 2020 Elsevier Ltd. All rights reserved.

Introduction

Over the past decade, wearable electronics have attracted sig-nificant attention around the world for their high flexibility, mul-tifunctionality, and continuous data measurement [1e8]. The on-going advances in engineering science include elaborate structuraldesigns [9e14] and innovative materials [15e17] that accelerate thedevelopment of flexible electronics integrated with intrinsicallyrigid functional materials. To meet the increasing demand for real-time monitoring of human health, many efforts have been made todevelop a comprehensive catalog of flexible health-care electronics,

), wenjli@cityu.edu.hk (W. Li),u).

aimed to eliminate the traditional rigid-interface platforms[18e21]. Among the various physiological human health parame-ters of interest, continuouslymonitoring the respiration rate/breathcan provide valuable medical data for human health assessments[22]. Currently, the types of respiration sensors that have been re-ported are mainly based on humidity-sensitive materials [23e27],pyroelectric materials [28,29], and triboelectric devices [30,31](Table S1). Although many respiration sensors based on humiditysensitivity declared their excellent performance at high relativeenvironmental humidity [24,32], the sensors tend to malfunctionas the ambient relative humidity approaches 100%. The respirationsensors that aim to take advantage of pyroelectric and triboelectriceffects to increase the accuracy of the measurements have bulkdesigns that limit their applications in daily life [28,31,33].Compared with the aforementioned respiration sensors, epidermalelectronics based on thermal convection effect for monitoring hu-man respiration were seldom investigated.

Although it remains a relatively new field of investigation, a fewresearchers have used flow sensors based on the thermal

Y. Liu et al. / Materials Today Physics 13 (2020) 1001992

convection effect with low power consumption [34,35]. The re-ported flow sensors exhibit a high-temperature resolution and canachieve extremely low air flow velocity measurement (<1 cm/s)[35], demonstrating their great potential in detecting humanrespiration. However, the rigid circuit platforms currently used ownmany limitations in the applications of wearable health-caremonitoring because they are not mechanically compatible with thecurvilinear morphology of the human body. To accurately detecthuman respiration, we demonstrate the capabilities of a highlyflexible, stretchable, and lightweight thermosensitive respirationsensor. The key functional component is a negative thermal coef-ficient thermistor, highly sensitive to the ambient temperature. Tofurther increase the temperature variation between the breathing(inhale or exhale) gas and the thermistor, a thermal actuator with afractal curve design distributes the heat around the thermistor,remarkably enhancing its sensitivity to inhaling (or exhaling) gas.Guided by theoretical modeling, the advanced structure designenables conformal integration with the skin under large mechani-cal deformations. By accurately detecting the five different respi-ration patterns, the sensor provides a new strategy for real-timemonitoring of human respiration.

Experimental section

Assembly of the respiration monitoring device

The fabrication process started on a quartz glass, which hadbeen cleaned by acetone, isopropanol, and deionized water (DIwater) sequentially. A thin film of poly(methyl methacrylate)(PMMA; thickness, ~200 nm) was spin coated onto the glass at2,000 rpm (solution concentration of 20 mg/ml) for 30 s and thenbaked on a hot plate at 200 �C for 20 min, serving as the sacrificiallayer. A thin layer of polyimide (PI; 2 mm) was formed on PMMA byspin coating (poly(amic acid) solution ¼ 12.0 wt % ± 0.5 wt %, at3,000 rpm for 30 s, baked at 250 �C for 30 min), serving as thesupporting layer. Next, Au/Cr (200/10 nm) was sputtered onto thePI film and then patterned by photolithography, yielding thin metaltraces of the designed geometries. Here, a positive photoresist (PR;AZ 5214; AZ Electronic Materials) was spin coated on the metal filmat 3,000 rpm for 30 s, soft baked on a hot plate at 110 �C for 4 min,then exposed to ultraviolet light for 5 s, and finally developed for15 s in a solution (AZ 300MIF). After metal etching, the PR wasremoved by acetone and rinsed by DI water. Then, another layer ofPI (2 mm)was spin casted on top of themetal traces at 3,000 rpm for30 s and annealed at 250 �C for 30 min. Selective dry etching of thetop PI layer was carried out by reaction ion etching (RIE; OxfordPlasma-Therm 790 system) at the power of 200 W for 10 min,forming the top encapsulation layer and exposing the area for theelectrodes. Here, selective etching was realized by protecting thepatterns defined by photolithography (AZ 4620 as the PR, AZElectronic Materials). The sample was immersed in acetone for 12 hto dissolve the PMMA sacrificial layer. Next, water soluble tapes(WSTs) were used as stamps to pick up the PI-encapsulatedmetallicelectrodes and sputtered with another thin layer of SiO2 (50 nm).Polydimethylsiloxane (PDMS; thickness of ~0.11 mm, elasticmodulus of ~145 kPa) substrates and the WSTs were exposed toultraviolet-induced ozone (UVO) for 10 min. The WSTs wereattached onto the PDMS substrate and then heated in an oven at70 �C for 10 min, forming strong chemical bonding between theelectrode and the substrate. Owning to the water-soluble charac-teristcis of the WSTs, immersing the sample in water for 5 min toremove the WSTs could realize soft stretchable electrodes. Next, aminiaturized negative temperature coefficient thermistor (ther-mal-sensitive constant, B ¼ 3786 K; resistance at temperature T,RT¼25�C ¼ 100 kU) was connected to the electrode patches using a

silver paste, then dried at 250�C for 3 min using a blower gun. It isfound that the device can still operate as original after over 1500stretching at a constant elongation of 10%. Finally, a thin layer ofPDMS (~0.11 mm, elastic modulus of 145 kPa) was spin coated ontop at 1,000 rpm for 30 s and cured at 110 �C for 5 min, serving asthe top encapsulation layer.

Mechanical modelling

The finite element analysis (FEA) commercial software ABAQUSwas used to design the fractal heating electrodes and the overalldevice layout for decreasing the strain level of the device whensubjected to different typical mechanical loads. The PDMS encap-sulation was modeled by hexahedron elements (C3D8R), whereasthe thin PI and Au layers weremodeled by shell elements (S4R). Theminimal element size in the model was one-fourth of the width ofthe Au wires (12.5 mm) to guarantee both the convergence andaccuracy of the simulation results. The elastic modulus and Pois-son's ratio used in the analysis were 78 GPa and 0.44 for Au,145 kPaand 0.5 for PDMS, 2.5 GPa and 0.34 for PI, and 119 GPa and 0.4 forthe thermistor, respectively.

Thermal modeling

The commercial software ABAQUS was used to study the tem-perature change of the mini thermistor as a function of the heatconvection coefficient for different input powers of the heatingelectrode. The entire device was modeled by hexahedron heat-transfer elements (DC3D8). Mesh convergence of the simulationwas ensured in all the cases. The thermal conductivity, heat ca-pacity, and mass density used in the simulations were1.0 W m�1 K�1, 2,200 J kg�1 K�1, and 840 kg m�3 for glass;130Wm�1 K�1, 490 J kg�1 K�1, and 6,150 kgm�3 for the thermistor;318 W m�1 K�1, 128 J kg�1 K�1, and 19,300 kg m�3 for Au;0.15 W m�1 K�1, 1,510 J kg�1 K�1, and 1,000 kg m�3 for PDMS; and0.21 W m�1 K�1, 2,100 J kg�1 K�1, and 909 kg m�3 for PI,respectively.

Characterization

The resistance data were collected using a data acquisition/multimeter system (DAQ6510, KEITHLEY). Thermal analysis wasconducted by capturing images with an infrared camera (FLIR C3).The airflow rate monitoring was calibrated using a gas mass flowmeter (MF5706-N-25, SIARGO Ltd.).

Results and discussion

Fig. 1a presents a schematic illustration of the fabrication pro-cess for a respiration sensor. To ensure surface smoothness andfabrication compatibility, a quartz glass was used as the substratefirst, with a spin-coated PMMA thin film on top as a sacrificial layerand a thin PI film as a metal electrodeesupporting layer. Metalliclayers (Cr/Au) were deposited on PI and patterned by photoli-thography, serving as the electrode for thermal actuators (Fig. S1,see Experimental section for details). Spin casting another PI andselectively dry etching by RIE provided a top encapsulation layer forthe thermal actuators and thus guarantee a reliable and robuststructure. Dissolving PMMAwith acetone enabled transfer printingof thermal actuators/circuits onto a PDMS substrate, which reliedon peeling off through a WST. To enhance the bonding strengthbetween the thermal electronics and the PDMS substrate, the sur-faces of both electronics attached WST and PDMS were treated byUVO for creating chemical bonds. The miniaturized thermistor(height of 0.6 mm, width of 0.3 mm, thickness of 0.23 mm) was

Fig. 1. Respiration sensor fabrication process and optical images. (a) Schematic illustration of the epidermal respiration sensor fabrication process. (b) Schematic diagram andoptical images of the respiration sensor with enlarged areas of the thermal actuator, the fractal electrodes, and the thermistor. (c) Optical image of the respiration sensor attached ontop of the fingertip. (d) Optical images of the respiration sensor mounted on top of the upper lip with two motions, including pouting (left) and compressing lips (right).PI ¼ polyimide; PMMA ¼ poly(methyl methacrylate); PDMS ¼ polydimethylsiloxane.

Y. Liu et al. / Materials Today Physics 13 (2020) 100199 3

mounted on the electrode patches using silver paste. The tiny ge-ometry of the thermistor provided great flexibility and stretch-ability of the whole device even under large mechanicaldeformations (Figs. S2 and S3). Finally, an additional layer of PDMSwas spin coated on top of the electronics to encapsulate theexposed electronic components. As per both experimental andtheoretical studies, the ultrathin PDMS encapsulation layer has anegligible effect on the thermal convection between humanrespiration (inhaling or exhaling) and the sensor but provides amechanically compliant and robust system. Fig. 1b presents theschematic illustrations and optical images of the stretchablerespiration sensor, wherein the enlarged areas show the detail ofthe electronic design and the sensor. The metal traces were used asboth electrodes and thermal actuators, wherein the fractal curveddesign enables its two-dimensional stretchability. The ultralowweight (~0.013 g) and compact overall size(8.3 mm � 6.3 mm � 0.22 mm) afford flexibility and comfortability

of use by a person. Fig. 1c and d show the optical images of thedevice mounted on the fingertip and upper lip of a human subject,demonstrating great flexibility and compatibility with human skin.The mechanical design, selection of filamentary fractal Au heatingelectrode patterns, and geometrical layouts of the entire devicewere guided by FEA in commercial software ABAQUS. Especially,the mechanical neutral plane of deformation for the thin Au elec-trodes between PI/PDMS layers minimizes the strain level duringmechanical loadings because the thickness of PI/PDMS at both sidesof the Au traces is the same. Therefore, the resulting deviceexhibited great flexibility and stretchability, which afforded acomfortable, non-irritating interface with the skin (Fig. S4). Thebreathability and air permeability of the sensor can be furtherimproved by creating arrays of small holes or using a breathablespandex substrate [20].

The optical images in Fig. 2a and the corresponding computeddistributions of strain of the Au layer in Fig. 2b show great

Fig. 2. Experimental and theoretical analysis of the respiration sensor. (a and b) Optical images and strain distribution (FEA) of a respiration sensor under stretching, bending, andtwisting deformations. (c) Electrical response of a respiration sensor to parallel air flow at 18.5 �C at a constant ambient temperature of 21 �C. (d) Temperature distributions of thedevice (marked by the boxed area) obtained by FEA and thermal imaging for a fixed airflow rate of 0.99 m/s. (e) Electrical response of the sensor intermittently flowing at a constantfrequency of 0.35 Hz under periodic air flow. (f) The optical images presenting the temperature distribution of the respiration sensor for both trough and peak cycle conditions asshown in (e). FEA ¼ finite element analysis.

Y. Liu et al. / Materials Today Physics 13 (2020) 1001994

mechanical properties of the respiration sensor. For instance, threetypical deformations of stretching, bending, and twisting werestudied. The equivalent strains remain below the elastic limit (0.3%)for ~18% stretching, ~180� bending (bending radius ~2 mm), and~90� twisting. These results highlight the range of robust andelastic responses, which is necessary to accommodate realisticphysiological motions. Considering the accuracy of the electricalsignal of the respiration sensor, a high signal-to-noise ratio is anessential point. Owing to the small temperature difference betweenthe area of the upper lip (~35 �C, Fig. S5) and exhaled air (~32 �C)[35,36], the resistance change of the sensor only exhibits a verysmall variation from 61 kU to 61.4 kU without additional amplifi-cation for recording respiration pattern of sitting (Fig. S6). A ther-mal amplifier associating with a heating actuator with a fractalcurve design (resistance of 714 kU) introduced here can distributethe heat around the thermistor and therefore increase the tem-perature gradient between the sensor and the exhaled air. As aresult, resistance change of the sensor is significantly amplified. For

instance, an extended resistance ranging from 43.7 to 44.7 kU canbe realized under the respiration pattern of sitting.

To find out the proper input power for the thermal actuator, weinvestigated the signal-to-noise ratio for various power inputs. Here,the device was fixed in an airtight box (environmental temperatureof 19 �C; relative humidity of 68%), with an artificial breathing pathcontrolled by a mass flow regulator, as shown in Fig. S7. For a con-stant heating power of 0.140 W and without any air flow, the tem-perature around the device stabilizes at 44.7 �C and the resistance ofthe thermistor (RS) reaches 49.6 U, as shown in Fig. S8. The RS in-creases linearly from 49.6 U to 66.4 U, with the square root of theairflow rate from 0 to 1.41 m1/2/s1/2(Fig. 2c). Similar trends can alsobe found in otherdifferentpower inputs. Another observation is thatthe sensitivity of the sensor can be significantly enhanced by higherinput power. The experimental results agree very well with thetheoretical model, indicating great reliability of the respirationsensor. The relationship between the resistance RS, airflow v, andtemperature of the thermistor is as follows [37]:

Fig. 3. Five respiration patterns recorded by the device. (a) Electrical signal measured by the respiration sensor under five respiration patterns, including panting, frightening,meditating, sitting, and sleeping. (b) Electrical signal amplitudes of the five respiration patterns.

Y. Liu et al. / Materials Today Physics 13 (2020) 100199 5

Rs ¼Rs0e�B

�1T0� 1

T*þPDTðvÞ

�(1)

where RS0 is the resistance of the thermistor at temperature T0, T* isthe thermistor temperature without heating, p is the input powerfor the thermal actuator, B is a constant value for a given thermistor,and DT(v) is the temperature change under the airflow withp ¼ 1 W. In our case, with an R0 of 100 kU, B of 3,786 K, T0 of298.15 K, and T* of 292.15 K, DT(v) is determined by FEA (Fig. S9),wherein the relationship between the heat convection coefficient hand v is as follows: h ¼ 13.6 v, and the unit is W/K m2 [38].

The sensitivity of the device can be improved significantly withhigher input power, as shown in Fig. 2c. The temperature of thedevice mounted on the upper lip is always lower than 40 �C duringnormal breath, but can reach up to ~52.5 �C with a 0.140-W heatingpower when the human subject suddenly holds his/her breath(Fig. S10). Here, the sudden rise of temperature can be alsoconsidered as another potential parameter for detecting anyrespiration arrest. However, considering the uncomfortablesensation of humans under high temperature for long terms, higherinput power should not be further considered even if it can improvethe sensitivity. Here, 0.140 W is adopted as the heating power toensure proper functionality while maintaining comfortable tem-perature sensation (<40 �C during breath) for the human subjects.The right panel in Fig. 2c presents the dynamic RS values of thedevice at a constant air flow rate of 0.99 m/s, demonstrating itselectrical stability during the operation. Fig. 2d compares thetheoretical and experimental results of temperature distributionfor the respiration sensor at a constant airflow rate of 0.99 m/s,showing good agreement between FEA and experiments. To further

investigate its stability, we measured the electrical signal of thedevice under various mechanical deformations, including static,stretching, bending, and twisting, and it was found that thesemechanical deformations had few impacts on its performance(Fig. S11). Next, we studied the dynamic response of the respirationsensor to the periodic on and off of a fixed air flow, as shown inFig. 2e and S5. The air flow rate was fixed at 0.99 m/s, on/off was setat a periodic time interval of 4.3 s (~0.35 Hz), and the respirationsensor can accurately capture the on and off of the air flow thatrevealed oscillatory changes in the electrical resistance from 57.0 to59.1 kU (highlighting the peak and trough in Fig. 2e). The temper-ature distributions of the device associating with peak and troughin Fig. 2e were obtained by infrared camera (IR) imaging at aminimum temperature of 35.2 �C and maximum temperature of40 �C, respectively (Fig. 2f). Fig. S12 presents the ambient temper-ature and relative humidity effect on the performance of the sensor.Obviously, the electrical signal decreases with the increase of theambient temperature ranging from 19 to 29 �C as the temperatureand airflow rate of the inflow gas are 19 �C and 0.99 m/s, respec-tively. In addition, the ambient relative humidity has a negligibleeffect on the signal response, possibly resulting from the excellentwaterproof characteristics of the encapsulation layers (PDMS).

Fig. 3a presents the measurement results of a volunteer for fivedifferent respiration modes, including gasping (2 Hz;57.6e58.7 kU), frightening (0.67 Hz; 46.4e47.7 kU), meditating(0.17 Hz; 47.8e54.0 kU), sitting (0.25 Hz; 43.7e44.7 kU), andsleeping (0.14 Hz; 37.4e38.4 kU), wherein the distance between thesensor and nostril is fixed at 3 mm. Evidently, each respirationmode has a unique signal response corresponding to differentamplitudes and frequencies which rely on the flow rate and fre-quency of exhaled breathing. For those low-frequency respiration

Fig. 4. Demonstrations of real-time monitoring of respiration. (a) Measurements of human respiration by the sensor during exercising and resting. The inset images show theexercising and resting states of the volunteer. (b) Optical images of real-time sleep monitoring of a volunteer, in which the sensor is mounted on the upper lip. (c) Dynamicmeasurement results of respiration by the sensor during the sleeping state of the volunteer.

Y. Liu et al. / Materials Today Physics 13 (2020) 1001996

modes, such as sitting, sleeping, andmeditating, the sensor was notonly able to fully detect both inhaling and exhaling processes (seeSupplement Video S1) but also able to clearly distinguish therespiration rate. Combination of this information can provideimportant clinical relevant information for sleep monitoring,intensive care unit, and many others. For the higher frequencyrespiration modes, such as gasping and frightening, the deviceexhibits higher sensitivity as RS variation is higher, as summarizedin Fig. 3b. Fig. S10 shows the electrical response as the humansubject suddenly holds breath, imitating the situation when a pa-tient in critical condition stops breathing unexpectedly. The

absence of thermal convection from respiration leads to a highertemperature around the thermistor, causing the resistance of thesensor to start decreasing. By including an additional alarm systemfor life-threatening scenarios, i.e., unexpected prolonged inter-ruption of breath, the respiration sensor can rise to become apromising human health monitoring choice in the health-carefields.

The following is the supplementary data to this article:VideoS1. To further investigate the device capabilities in clinicallyrelevant respiration monitoring, the respiration status of a volun-teer under physical exercise and rest was measured. Fig. 4a shows

Y. Liu et al. / Materials Today Physics 13 (2020) 100199 7

the real-time respiration signal measured by the device duringexercising (~0.5 round per minute) and rest. The fast respirationrate (~0.4 Hz) caused by the strenuous exercise significantly con-tributes to the thermal convection between the device andbreathing gas, leading to a relatively high RS response, ranging from47.9 to 51.6 kU. Moreover, large signal amplitude fluctuationsreflect the random respiration and breathing air flow rates of thehuman subject during exercising. During the break, the decreasingrespiration rate (~0.34 Hz on average) contributes to a lower ther-mal convection, leading to the electrical signal response eventuallyapproaching to the sitting respiration mode previously shown inFig. 3a. Next, sleep monitoring by the epidermal sensor wasdemonstrated as shown in Fig. 4b, in which the optical imagesshow a volunteer turning his head during a nap and thus imitatethree sleeping statuses of static, 45� head turning, and 90� headturning. Three obvious fluctuations of the RS happened during thenap, and the real-time monitoring of the sensor clearly recordedthese signals, demonstrating that the device can accuratelymonitor the body motions during sleeping, potentially providing aquantitative approach to record a patient's sleep history overnight.During the static head mode, the RS ranges from 37.5 to 38.5 kU,similar to the sleeping respiration result in Fig. 3a. As a result of theincreasing gas flow and headmotions during sleep, the heat aroundthe device significantly dissipates, resulting in an evident increaseof RS, as shown in Fig. 4c. In addition, it was found thatmore intensehead motions contribute to greater signal fluctuations, 41.7 kU and47.9 kU at the peak for turning the head 45� and 90� are detected,respectively.

Conclusion

In summary, a skin-integrated, flexible, and stretchable respi-ration sensor based on thermal convection effect is developed tomonitor human respiration in real time. By introducing a thermalactuator consisting of fractal structure metal traces, the sensorbecomes highly sensitive to human respiration at a broad range offrequency. The nature of the thermal convection effect allows themeasured electrical signal amplitudes and frequencies to be easilydistinguished under different respiration modes and thus proveseasily analyzed and relevant clinical standard data. In addition,real-time monitoring of the respiration rate/depth during periodsof physical exercise, rests, and prolonged rest (siesta) demonstratesthe feasibility of the sensor to be featured in future diagnosticbreath analysis and respiration status recognition.

Declaration of competing interest

No competing interests.

Acknowledgments

This work was supported by City University of Hong Kong (grantno. 9610423).

Appendix A. Supplementary data

Supplementary data to this article can be found online athttps://doi.org/10.1016/j.mtphys.2020.100199.

References

[1] C. Dagdeviren, P. Joe, O.L. Tuzman, et al., Recent progress in flexible andstretchable piezoelectric devices for mechanical energy harvesting, sensingand actuation, Extreme mech. lett. 9 (2016) 269e281.

[2] Y.J. Hong, H. Jeong, K.W. Cho, N. Lu, D.H. Kim, Wearable and implantabledevices for cardiovascular healthcare: from monitoring to therapy based on

flexible and stretchable electronics, Adv. Funct. Mater. 29 (19) (2019)1808247.

[3] Z. Xie, R. Avila, Y. Huang, J.A. Rogers, Flexible and stretchable antennas forbiointegrated electronics, Adv. Mater. (2019) 1902767.

[4] H.U. Chung, B.H. Kim, J.Y. Lee, et al., Binodal, wireless epidermal electronicsystems with in-sensor analytics for neonatal intensive care, Science 363(6430) (2019): eaau0780.

[5] X. Yu, H. Wang, X. Ning, et al., Needle-shaped ultrathin piezoelectric micro-system for guided tissue targeting via mechanical sensing, Nat. Biomed. Eng. 2(3) (2018) 165.

[6] Y. Liu, L. Zhao, L. Wang, et al., Skin-integrated graphene-embedded leadzirconate titanate rubber for energy harvesting and mechanical sensing, Adv.Mater. Technol. 4 (12) (2019) 1900744.

[7] J.C. Yang, J. Mun, S.Y. Kwon, S. Park, Z. Bao, S. Park, Electronic skin: recentprogress and future prospects for skin-attachable devices for health moni-toring, robotics, and prosthetics, Adv. Mater. 31 (48) (2019) 1904765.

[8] S.H. Jeong, A. Hagman, K. Hjort, M. Jobs, J. Sundqvist, Z. Wu, Liquid alloyprinting of microfluidic stretchable electronics, Lab Chip 12 (22) (2012)4657e4664.

[9] Y. Liu, Y. Xu, R. Avila, et al., 3D printed microstructures for flexible electronicdevices, Nanotechnology 30 (41) (2019) 414001.

[10] P. Gutruf, V. Krishnamurthi, A. V�azquez-Guardado, et al., Fully implantableoptoelectronic systems for battery-free, multimodal operation in neurosci-ence research, Nat. Electron. 1 (12) (2018) 652.

[11] C. Pang, J.H. Koo, A. Nguyen, et al., Highly skin-conformal microhairy sensorfor pulse signal amplification, Adv. Mater. 27 (4) (2015) 634e640.

[12] Y. Liu, L. Wang, L. Zhao, et al., Thin, skin-integrated, stretchable triboelectricnanogenerators for tactile sensing, Adv. Electron. Mater. 6 (1) (2020).

[13] Y. Yu, H. Yuk, G.A. Parada, et al., Multifunctional “hydrogel skins” on diversepolymers with arbitrary shapes, Adv. Mater. 31 (7) (2019) 1807101.

[14] K. Sim, S. Chen, Z. Li, et al., Three-dimensional curvy electronics created usingconformal additive stamp printing, Nat. Electron. 2 (10) (2019) 471e479.

[15] X. Yu, L. Zeng, N. Zhou, et al., Ultra-Flexible,“Invisible” thin-film transistorsenabled by amorphous metal oxide/polymer channel layer blends, Adv. Mater.27 (14) (2015) 2390e2399.

[16] X. Yu, T.J. Marks, A. Facchetti, Metal oxides for optoelectronic applications,Nat. Mater. 15 (4) (2016) 383.

[17] S. Huang, Y. Liu, Y. Zhao, Z. Ren, C.F. Guo, Flexible electronics: stretchableelectrodes and their future, Adv. Funct. Mater. 29 (6) (2019) 1805924.

[18] L. Tian, B. Zimmerman, A. Akhtar, et al., Large-area MRI-compatible epidermalelectronic interfaces for prosthetic control and cognitive monitoring, Nat.Biomed. Eng. 3 (3) (2019) 194.

[19] Z. Huang, Y. Hao, Y. Li, et al., Three-dimensional integrated stretchable elec-tronics, Nat. Electron. 1 (8) (2018) 473.

[20] X. Yu, Z. Xie, Y. Yu, et al., Skin-integrated wireless haptic interfaces for virtualand augmented reality, Nature 575 (7783) (2019) 473e479.

[21] Z. Wu, S. Zhang, A. Vorobyev, et al., Seamless modulus gradient structures forhighly resilient, stretchable system integration, Mater. Today Phys. 4 (2018)28e35.

[22] C.-K. Peng, J.E. Mietus, Y. Liu, et al., Quantifying fractal dynamics of humanrespiration: age and gender effects, Ann. Biomed. Eng. 30 (5) (2002) 683e692.

[23] D. Zhang, Y. Cao, P. Li, J. Wu, X. Zong, Humidity-sensing performance of layer-by-layer self-assembled tungsten disulfide/tin dioxide nanocomposite,Sensor. Actuator. B Chem. 265 (2018) 529e538.

[24] B. Li, Q. Tian, H. Su, X. Wang, T. Wang, D. Zhang, High sensitivity portablecapacitive humidity sensor based on In2O3 nanocubes-decorated GO nano-sheets and its wearable application in respiration detection, Sensor. Actuator.B Chem. 299 (2019) 126973.

[25] J. Dai, H. Zhao, X. Lin, et al., Ultrafast response polyelectrolyte humidity sensorfor respiration monitoring, ACS Appl. Mater. Interfaces 11 (6) (2019)6483e6490.

[26] X. Huang, B. Li, L. Wang, X. Lai, H.-G. Xue, J-f Gao, A superhydrophilic, un-derwater superoleophobic and highly stretchable humidity and chemicalvapor sensor for human breath detection, ACS Appl. Mater. Interfaces 11 (27)(2019).

[27] Y. Pang, J. Jian, T. Tu, et al., Wearable humidity sensor based on porous gra-phene network for respiration monitoring, Biosens. Bioelectron. 116 (2018)123e129.

[28] H. Xue, Q. Yang, D. Wang, et al., A wearable pyroelectric nanogenerator andself-powered breathing sensor, Nanomater. Energy 38 (2017) 147e154.

[29] K. Roy, S.K. Ghosh, A. Sultana, et al., A self-powered wearable pressure sensorand pyroelectric breathing sensor based on GO interfaced PVDF nanofibers,ACS Appl. Nano Mater. 2 (4) (2019) 2013e2025.

[30] S. Wang, H. Tai, B. Liu, et al., A facile respiration-driven triboelectric nano-generator for multifunctional respiratory monitoring, Nanomater. Energy 58(2019) 312e321.

[31] H. Zhang, J. Zhang, Z. Hu, et al., Waist-wearable wireless respiration sensorbased on triboelectric effect, Nanomater. Energy 59 (2019) 75e83.

[32] Y. Lee, S.H. Cha, Y.W. Kim, D. Choi, J.Y. Sun, Transparent and attachable ioniccommunicators based on self-cleanable triboelectric nanogenerators, Nat.Commun. 9 (2018) 1e8.

[33] S. Wang, Y. Jiang, H. Tai, et al., An integrated flexible self-powered wearablerespiration sensor, Nanomater. Energy 63 (2019) 2211e2855.

[34] P. Liu, R. Zhu, R. Que, A flexible flow sensor system and its characteristics forfluid mechanics measurements, Sensors 9 (12) (2009) 9533e9543.

Y. Liu et al. / Materials Today Physics 13 (2020) 1001998

[35] A.S. Cubukcu, E. Zernickel, U. Buerklin, G.A. Urban, A 2D thermal flow sensorwith sub-mW power consumption, Sensor Actuator Phys. 163 (2) (2010)449e456.

[36] Y.J. Choi, S.H. Min, J.J. Park, J.E. Cho, S.Z. Yoon, S.M. Yoon, Comparison of thetemperature and humidity in the anesthetic breathing circuit among differentanesthetic workstations: updated guidelines for reporting parallel grouprandomized trials, Medicine 96 (25) (2017).

[37] A. Feteira, Negative temperature coefficient resistance (NTCR) ceramicthermistors: an industrial perspective, J. Am. Ceram. Soc. 92 (5) (2009)967e983.

[38] S. Gao, R. Ooka, W. Oh, Formulation of Human Body Heat Transfer Coefficientunder Various Ambient Temperature, Air Speed and Direction Based on Ex-periments and CFD, Building and Environment, 2019, p. 106168.