A Robust Bridge‐Type Airflow Sensor Based on Flexible Superhydrophobic Carbon Nanotube Fiber Thin Films

Despite a great deal of research about airflow sensors by far, it is still hard to maintain ultra‐low limit of detection, short response, and recovery time under humid and rainy environment, which compromises its practical applications. Herein, a robust bridge‐type airflow sensor base on flexible superhydrophobic carbon nanotube fiber thin films is developed, which can not only own remarkable sensitive properties with ultra‐low limit of detection, fast response, and rapid recovery, but also exclude the influence of moisture or water drops in real applications. Both response time and recovery time are reduced compared with that of the previously reported flexible carbon‐based airflow sensors. Particularly, recovery time is reduced by ≈2/3. Meanwhile, the superhydrophobic structure renders normal functioning of the sensor in severe environment of high humidity or even in the rain, so that the influence of moisture and raindrops can be excluded, a property deficient in the previously reported airflow sensors. Not only can the sensor detect ambient airflow sensitively, but it also exhibits outstanding performance in respiratory health monitoring. Hence, the novel sensor designed in this work realizes operation in the environment of high humidity and rain, which tremendously uplifts the applicability of flexible sensors.


Introduction
With the rapid development of science and technology and intelligence, all kinds of sensors, including temperature, [1,2] humidity, [3][4][5] air flow, [6][7][8][9] have developed explosively and been widely used in people's life and production, playing a crucial role in promoting social development.On the other hand, the fast-progressing society continues to throw higher demands on DOI: 10.1002/admi.202300077sensors for better performance and applicability in harsh environments.Sensors also undergo huge transition in material composition from the once predominant rigid semiconductor sensors toward flexible sensors using polymers and nanomaterials as sensitive elements. [10,11]][20][21] Most traditional airflow sensors utilize redox reactions between some gases in the air and semiconducting oxide sensitive elements to alter resistance of the sensitive elements and hence realize airflow detection. [22]In other cases, strain effect [23] and piezoelectric effect [24,25] mechanisms are also adopted for airflow sensing.Drawbacks of retarded response and recovery, narrow detection range, high energy consumption, and limited applicable scenarios severely restrict the application of traditional airflow sensors.In recent years, high performance materials have been continuously developed, among that the emergence of carbon nanotubes leads to extensive report on carbon nanotube airflow sensors.Studies on gas sensitivity of carbon nanotubes can be dated back to 2001, [12] after which airflow sensors based on carbon nanotubes and corresponding composite structures attracted wide research interest.A previously reported hot wire flow sensor adopted carbon nanotubes with Al 2 O 3 coatings as the sensitive elements, exhibiting good sensibility and demonstrating the application prospect of carbon nanotubes in airflow sensors, but its response speed and recovery speed are slow. [26]ecently, an airflow sensor based on suspended carbon nanotube networks was reported, showing rapid response and good sensitivity, However, the silicon substrates of the sensor render incompatibility with flexible wearable smart devices, while the nano networks comprising of carbon nanotubes account for narrow detection range, long recovery time, and difficulty in operation and practicality. [27]Meanwhile, there have been massive research and reports on bionic airflow sensors in the past few years, spanning from mimicry of artificial hair, animal fur, and whiskers at the early stage, [28][29][30] to a recently reported carbon nanotube airflow sensors that suffered from low response and recovery speed. [31]evertheless, carbon nanotube airflow sensors are still under experimental study at present, meaning that other environmental factors, especially water vapor, are not taken into consideration in terms of the sensing properties of sensors.Water vapor always exerts influence on the sensitive elements of most sensors, especially airflow sensors, as most airflow carries water vapor that significantly affects conductivity of carbon nanotubes. [32]At the same time, most carbon nanotube airflow sensors adopt single carbon nanotubes as the sensitive elements, so that the narrow detection range, greatly varied response and recovery time, and monotonous application scenario make for the difficulty in industrialized production and application of these sensors.
To tackle the current problems faced by carbon nanotube airflow sensors, we were inspired by bridge structures and the fact that the bridge decks are subjected to different forces when facing winds from different directions and developed a flexible bridgetype airflow sensor based on super-hydrophobic carbon nanotube fiber thin films.The sensor was competent in airflow detection in the direction range of 0°-180°.Carbon nanotube fiber thin films are ultra-thin networks composed of multi-wall carbon nanotubes.Reversible on/off switch of airflow will change the contact spots among tubes in carbon nanotube fiber thin films, thus altering conductivity of carbon nanotube fiber thin films.Network carbon nanotube fiber thin films composed of nano-level carbon nanotubes exhibit good flexibility and elasticity, endowing the airflow sensor with remarkable performance such as ultralow limit of detection (≈0.1 m s −1 ), fast response (≈0.2 s), and fast recovery (≈0.35 s).Additionally, the sensitive elements feature with super-hydrophobic structures bearing superhydrophobicity, self-cleaning, and anti-fouling properties, which make it possible for the application of the sensor in harsh environments of high humidity, rain, and acidity.This is the property most airflow sensors lack at present.Meanwhile, the sensor also exhibits ultrahigh sensitivity, stability, instantaneity, and rapidity in respiratory health monitoring, making the sensor not only suitable for airflow detection but also breathing detection to monitor cardiopulmonary function and sleep quality, as well as breathing recognition, hence showing extensive application prospect of the sensor.

Preparation of Carbon Nanotube Fiber Thin Film Super-Hydrophobic Structure
Figure 1a shows the preparation process of carbon nanotube fiber thin film (CNTF) superhydrophobic structures.Coarse structures were constructed on the surface of CNTF, while the adopted CNTF was synthesized through floating catalyst chemical vapor deposition (FCCVD). [33]The optical image, SEM image and EDX result were illustrated in (Figure 1b,c) showing the network structure of CNTF.Preparation of superhydrophobic coarse structures was carried out following 3 steps, starting with synthesis of superhydrophobic composites.At room temperature, 0.1 mol L −1 aqueous solution of sodium perfluorooctanoate (40 mL) (SPFO) was added drop by drop while stirring into 1.0 mg mL −1 aqueous solution of poly dimethyl diallyl ammonium chloride (200 mL) (PDDA), in which nano SiO2 (1.2 g) was dispersed by ultrasound.After drying the suspension in an electro-thermal blowing dry box and grinding the dried solids in a mortar, PDDA-PFO/SiO2 hydrophobic composites was obtained.The hydrophobicity of the hydrophobic composites mainly originated from the surface abundant fluoride-containing hydrophobic groups and some hydrophilic quaternary ammonium groups of the amphiphilic substance PDDA-PFO generated by reaction between PDDA and SPFO (Figure S1, Supporting Information).Major functions of SiO2 were binding with hydrophilic quaternary ammonium groups on PDDA-PFO using the abundant hydrophilic hydroxy groups on the surface of SiO2 on the one hand and serving as the matrix for PDDA-PFO on the other hand.
Subsequently, CNTF coarse structures were constructed.First, polyvinyl alcohol (1 g) (PVA) was added into deionized water (1000 mL), followed by 3 h of heating in a 90 °C water bath for complete dissolution to yield 0.1% PVA aqueous solution.Then, SiO 2 (1 g) nanoparticles were added into the prepared 0.1% PVA (100 mL) aqueous solution to generate PVA-SiO 2 mixture after 2 h of ultrasound.CNTF was soaked in the PVA-SiO 2 .mixturefor 1 h.Coarse structures on the surface of CNTF formed by PVA and SiO 2 were achieved after drying, which were denoted as CNTF-PVA-SiO 2 -1.The reasons for using PVA and SiO 2 to prepare coarse structures were as follows.PVA was a good softner and binder, with superb gas barrier property [34,35] and bipolarity (PVA structure was shown Figure S2, Supporting Information).Due to the co-existing hydrophilic hydroxy groups and hydrophobic vinyl acetate monomers in PVA, it can strongly bind CNTF and endow CNTF-PVA composites with hydrophilicity, while combining with SiO 2 through abundant hydroxy groups and strong adhesion, so that the obtained coarse structures were not just uniform in morphology, but also readily connected with hydrophobic composites.
Finally, CNTF super-hydrophobic structures were fabricated.PDDA-PFO/SiO 2 (0.1 g SiO 2 was added) composites were dispersed into anhydrous alcohol (10 mL), and the solution was uniformly coating on CNTF-PVA-SiO 2 -1 coarse structures.Superhydrophobic CNTF was obtained after drying, which was noted as CNTF-PPPS-1.The physical image, SEM images and EDX result are exhibited in Figure 1e,f and Figure S3c,d (Supporting Information), the microscopic coarse structures on CNTF surface and the compositional elements of CNTF-PPPS-1 composite structures.
To further investigate the influence of SiO 2 content on CNTF hydrophobicity and airflow sensors, contrast materials were prepared following the above-mentioned protocols except that 0.5 and 2 g SiO 2 nanoparticles were added into PVA solutions and were denoted as CNTF-PPPS-0.5 and CNTF-PPPS-2, respectively.(Figure S3e-g, Supporting Information) are SEM images of CNTF-PPPS-0.5,CNTF-PPPS-1, and CNTF-PPPS-2.By comparison, CNTF-PPPS-1 exhibited the best uniformity of surface coarse structures with apparent concave-convex morphology, while CNTF-PPPS-0.5 showed extremely nonuniform distribution of surface coarse structures that were absent on most CNTF surface.Multilayer surface coarse structures appeared in CNTF-PPPS-2, but the concave-convex morphology was irrecognizable.

Fabrication of Flexible Bridge-Type Airflow Sensors Based on Super-Hydrophobic Carbon Nanotube Fiber Thin Films
Fabrication process of flexible bridge-type airflow sensors based on super-hydrophobic carbon nanotube fiber thin films (Figure S4, Supporting Information), The principal sensing unit of the sensor is CNTF-PPPS sensitive elements, which possess remarkable performance of super-hydrophobicity, ultrathinness, and superb flexibility, it also has the ability to quickly recover from large deformations (Movie S2, Supporting Information).The substrate of the sensor was a bridge structure constituted by flexible acrylic transparent double-sided tape (AATDSA).Sensitive elements related to the substrate by copper foil electrodes.The waterproof property, high temperature resistance, flexibility, ductility, and robust adhesion of flexible AATDSA make the sensor readily compatible with different surfaces such as clothing, skin, and walls to carry out airflow detection after adhesion.Figure 1f shows the bridge-type super-hydrophobic airflow sensor inspired by bridge structures and the corresponding working principles.The impact of airflows induces distortion of CNTF-PPPS, and thereby alters its conductivity.Figure 1g and h demonstrate performance of CNTF-PPPS airflow sensors in airflow detection and breathing detection, respectively, showing fast response and recovery, and superb sensitivity of the sensor.

Breathing and Finger Non-Contact Experimental Operation Process
This experiment was carried out at a constant temperature of 23 °C.The CNTF-PPPS-1 airflow sensor is related to the digital meter Keithley 2450 to construct a sensing detection system.When carrying out the breathing experiment, people's breathing nostril was 3 cm away from CNTF-PPPS-1 airflow sensor, and then they breathe.With the change of exhalation and inhalation, the digital table Keithley 2450 dynamically displays the resistance change curve of the sensor, and the breathing state can be analyzed through the resistance change curve of the sensor.When the finger non-contact movement experiment was carried out, the finger was slowly moved up and down from 0.5 to 1.5 cm above the CNTF-PPPS-1 airflow sensor.With the change of the distance between the finger and the sensor, the digital table Keithley 2450 dynamically shows the sensor resistance change curve.This breathing and finger non-contact were done by the first author, these experiments do not disobey ethics committee approval.

Characterization and Study on Super-Hydrophobicity
Generally, contact angle and sliding angle are two parameters to measure wettability of solid materials.As shown in Figure S5a,b, contact angle is the angle measured through the liquid from solid-liquid interface to gas-liquid interface at the intersection of solid, liquid, and gas phases, which is the angle between  gl and  sl , usually noted as .The value of  ranges (0°-180°), reflecting the wettability of a solid by a liquid.Higher  represents poorer wetting ability of a liquid to a solid.Four kinds of surfaces with varied wettability are defined through the boundaries of 30°, 90°and 150°, namely super-hydrophilic, hydrophilic, hydrophobic, and super-hydrophobic.Sliding angle refers to the critical angle between the horizontal plane and a tilted surface, on which liquid droplets begin to slide, usually noted as  (As shown in Figure S5b, Supporting Information), small critical angle normally equals to poor adhesion between liquids and solid surfaces. [36]Figure 2a-d are optical images of water droplets on CNTF, CNTF-PPPS-0.5,CNTF-PPPS-1, and CNTF-PPPS-2 surfaces, respectively.Water droplets on the surfaces of CNTF and CNTF-PPPS-0.5 exhibited an elliptic sphere shape, while those on CNTF-PPPS-1, and CNTF-PPPS-2 appeared to be spherical, suggesting hydrophobicity of CNTF and CNTF-PPPS-0.5 surfaces and super-hydrophobicity of CNTF-PPPS-1, and CNTF-PPPS-2.Contact angle test results in Figure 2e-h further confirmed this conclusion.Super-hydrophobicity of CNTF-PPPS-1 surface can be explained from two aspects.First and the most important reason for the realization of super-hydrophobicity is the coarse structures formed by SiO 2 and PVA on CNTF surface.The uniform coarse structures impeded infiltration of liquid droplets into the cracks of the structure, so that an air layer can exist between liquid droplets and the surface of CNTF-PPPS-1.Therefore, the real contact angle phase was a mixed gas-liquid and liquid-solid contact phase (As shown in Figure S5c, Supporting Information).The second explanation is the PDDA-PFO generated by reaction between PDDA and SPFO (reaction formula is shown as Figure S2, Supporting Information).PDDA-PFO carried plentiful of hydrophobic fluoride-containing groups that modified hydrophobicity and adhesiveness of the micro sphere surface, leading to super-hydrophobicity and low adhesiveness of CNTF-PPPS-1 surface.Figure 2i-l are the physical images of CNTF, CNTF-PPPS-0.5,CNTF-PPPS-1, and CNTF-PPPS-2, respectively.According to the images, flexibility of CNTF-PPPS-0.5 and CNTF-PPPS-1 were slightly higher than that of CNTF, which was accounted for the softening effect of PVA.However, as SiO 2 content in the coarse structures increased, the added thickness compromised the flexibility of CNTF-PPPS-2.

Airflow Sensing Characteristics and Mechanisms
The fabricated flexible bridge-type airflow sensors featured ultrahigh sensitivity, fast response and recovery, multi-angle response, and wide detection range.CNTF-PPPS-0.5,CNTF-PPPS-1, and CNTF-PPPS-2 airflow sensors were prepared to investigate the influence of SiO 2 content in coarse structures on sensing properties of sensors.Figure 3a shows airflows with differed directions detected by airflow sensors.The angle between airflow direction and the sensor was noted by .Variation curve of relative resistance change of CNTF, CNTF-PPPS-0.5,CNTF-PPPS-1, and CNTF-PPPS-2 airflow sensors against airflow rate with the angle between airflow direction and the sensors set as 90°is shown in Figure 3a.Sensitivity of sensors is defined as the slope of resistance changing with airflow velocity ((R-R 0 )/R 0 (%)), where R and R 0 .arereal-time resistance under airflow and resistance without airflow, respectively.By conducting repeated sensing response tests on sensors prepared based on the four sensitive elements mentioned above, as well as conducting extensive sensing response tests on sensors prepared based on the same sensitive element, the variation curve of the absolute value of sensitivity against airflow intensity exhibited in Figure 3b was obtained.Sensitivity of CNTF, CNTF-PPPS-0.5,CNTF-PPPS-1, and CNTF-PPPS-2 sensors increased simultaneously with airflow intensity.And the CNTF-PPPS-1 airflow sensor showed higher sensitivity than that of CNTF-PPPS-0.5 and CNTF-PPPS-2, which was accounted for uniform surface coarse structures and better flexibility of the composite film.More uniform surface coarse structures and better flexibility contribute to higher sensitivity of sensors.Sensitivity of CNTF-PPPS-0.5 and CNTF-PPPS-1 airflow sensors was slightly higher than that of CNTF airflow sensors, which was mainly resulted from the enhanced flexibility and gas barrier property of the composite structures induced by PVA soaking.Meanwhile, coarse structures on the surface enlarged contact areas with airflow, making for higher sensor sensitivity.CNTF-PPPS-2 sensors exhibited the poorest sensitivity, since the surface coarse structures were too thick that the flexibility was severely compromised, thereby influencing sensitivity.At the same time, according to the curve of Figure 3b, the prepared different sensors show good stability and reproducibility, although some errors will occur in many repeated experiments, but the errors are less than 5%.The sensor can have excellent reproducibility and stability, mainly because it uses carbon nanotube fiber film as the basic conductive skeleton, and it has excellent conductivity, stability, flexibility, and deformation self-recovery ability.Figure 3b and c are the response time curve and recovery time curve under different airflows of CNTF, CNTF-PPPS-0.5,CNTF-PPPS-1, and CNTF-PPPS-2 airflow sensors, respectively.Response and recovery time of CNTF, CNTF-PPPS-0.5, and CNTF-PPPS-1 were approximate, among those CNTF showed slightly shorter response and recovery time, while that of CNTF-PPPS-2 was the longest.We also investigated sensor sensitivity curves of the CNTF and CNTF-PPPS airflow sensors upon on/off switch of 0.1-4 m s −1 airflow and obtained agreeing results of sensitivity, and response and recovery time as the above (Figure S6, Supporting Information).To better study other properties of super-hydrophobic airflow sensors, we selected CNTF-PPPS-1 sensors with higher sensitivity as the research subject in the following section.The on/off sensitivity curve of the CNTF-PPPS-1 airflow sensor against 0.1-4 m s −1 airflow shown in Figure 3e also proved high sensitivity and rapid response and recovery of the sensor against airflows with different velocity.To further study the angle  for the sensor to detect airflow directions, we investigated sensitivity of the CNTF-PPPS-1 airflow sensor subjected to airflows with different velocity within the range (0°-180°).Results show that sensor sensitivity was symmetrical along 90°(Figure S7, Supporting Information), as 0°-90 o angle between airflow and the sensor is the same as 90°-180 o angle but with different reference coordinate.Therefore, we investigated the sensor within 0°-90°angle between airflow and the sensor based on CNTF-PPPS-1 (Figure S8a, Supporting Information).As shown in Figure 3f, sensor sensitivity increased with , Particularly, the sensor also made response when  was set to 0. Additionally, the CNTF-PPPS-1 sensor exhibited great stability, reliability, and rapid response and recovery as shown in Figure 3g and h. Figure 3g show that when  was 0°, 45°and 90°, resistance of the CNTF-PPPS-1 sensor decreased immediately once subjected to airflow.When airflow was turned off, resistance of the sensor would soon recover to the initial state.Through regularly turning on and off 2 m s −1 airflow and constantly switching airflow direction, sensor resistance exhibited regular increase and decrease, demonstrating excellent on/off property and stability of the sensor, which was attributed to good flexibility and self-recovery of the ultra-thin CNTF-PPPS-1 films.We also studied the variation curve of sensor sensitivity against on and off airflow when subjected to different airflow (0.1-4 m s −1 ) at different angles ( = 0°, 45°and 90°).The rapid response, fast recovery, and the on/off property were consistent with the results above (Figure S8c-k, Supporting Information).As shown in Figure 3h, response time was only 0.2 s when the resistance changes of the airflow sensor reached 90%, while recovery time was only 0.35 s.Among all carbon-based material airflow sensors, this airflow sensor stands out as one of the airflow sensors with shortest response and recovery time at present.Currently, there are no carbon-based airflow sensors of which the response time and the recovery time are both below 0.4 s.According to Figure 3g and h, under the same airflow velocity, changes in sensitivity and resistance of airflow sensors were correlated to , mainly because the angle (0°-90°) between airflow direction and the sensitive elements of the sensor would influence the contact area of airflow and the sensor and the actual impact force of airflow on the sensor.When 0°<  ≤90°, major influence came from the actual impact force of airflow on the sensor.In the mechanical model of the impact from airflow on the sensor (Figure S8b, Supporting Information), the hypotenuse of the triangle refers to the direction of airflow, and the corresponding pressure was marked as F gas , while the right-angle side was the actual impact force of airflow on the sensor, namely F C-P .The angle between the hypotenuse and CNTF-PPPS-1 sensor  was the angle between airflow direction and the sensor.Therefore, applying mechanical triangle model, it can be deduced that the actual airflow impact force subjected to the sensitive elements of the sensor F C-P = F gas sin (0°<  ≤90°), meaning that sensor sensitivity increases simultaneously with  (0°<  ≤90°).When  = 0°, the airflow is parallel to the sensor.At this point, the airflow blows to the side of the sensor sensitive element (as shown in the dashed box in Figure S8a, Supporting Information), causing deformation on the sensor side to achieve sensor response.However, due to the thin thickness of the sensor sensitive element, the contact area between the airflow and the sensor is the smallest, and the deformation area of the sensitive element is small, resulting in the lowest sensitivity.Figure 3i shows the consecutive circulating signals of 250 times of on/off airflow from the sensor.The shape and the amplitude of the obtained curves remained basically unchanged, further confirming the stability, reliability, and good on/off characteristics of the sensor.As shown in Figure 3j, from the comparison of detection range and response time between CNTF-PPPS-1 sensor and the previously reported sensors, it can be analyzed that the sensor achieved both a low limit of detection of 0.1 m s −1 and a rapid response time of 0.2 s.Such low limit of detection and rapid response are hardly achieved by airflow sensors at present, which severely restricts application of these sensors such as CNTs/CSF airflow sensors, of which the lowest limit of detection is 0.005 m s −1 but the response time is 1.3 s. [31] Out of the remarkable comprehensive properties, the CNTF-PPPS-1 sensor can be applied for airflow detection in various environments and industrial production.
In order to better prove the excellent performance of the superhydrophobic CNTF-PPPS-1 air flow sensor that can still be used in rainy days and high humidity environments, CNTF-PPPS-1 and CNTF air flow sensors are simulated to detect air flow in rainy days and high humidity environments by dropping water on the superhydrophobic CNTF-PPPS-1 and hydrophobic CNTF air flow sensors when detecting air flow, and see the changes in the R-T curve of CNTF-PPPS-1 and CNTF air flow sensors before and after dropping water.Drip model diagram of sensor air flow detection (as shown in Figure S9a, Supporting Information). Represents the angle between the sensor and the horizontal plane, while  is the angle between airflow direction and the sensor.Airflow was fixed along the horizontal direction, whereas the direction of water droplets was fixed perpendicularly to the horizontal plane.Then we change the size of  ( =  is always equal), to verify the effect of water droplet on the performance of the sensor at different tilt angles, we mainly studied  is 15 o , 45 o and 90 o respectively (Movie S2, Supporting Information).Figure S9b and c (Supporting Information) are the models of the movement of water droplets on the sensitive element surfaces of CNTF-PPPS-1 and CNTF sensors, respectively.Figure 4a-c are the optical overlap diagrams of the movement of water droplets on CNTF surface with being 15°, 45°and 90°, respectively.Once contacting the surface of CNTF, water droplets were stuck and could not detach from the surface.Figure 4d-f   (enlarged image of the dashed box in (g)).i) Sensing curve of 600 times of airflow on/off switch cycles measured by the CNTF-PPPS-1 sensor at a airflow velocity of 2 m s −1 ,  = 90°.26,[41][42][43][44][45] (Note: The negative sign in (R-R 0 )/R 0 value represents only the direction of resistance change, while negative sign indicates the decrease of resistance) the surface of CNTF-PPPS-1, water droplets bounced back and separated from the surface immediately.The information above articulates the macroscopic phenomena of the movement of water droplets on the surface of CNTF and CNTF-PPPS-1.To better investigate the influence of water droplets on the performance of CNTF and CNTF-PPPS-1 airflow sensors, we carried out water dropping tests while airflow detection of sensors.Figure 4g-i are the R-T curves of the CNTF airflow sensor subjected to 5 μL water droplets on the surface at an airflow velocity of 2 s −1 with  being 15°, 45°and 90°, respectively.According to the measured R-T curves, when  = 15°and 45°, sensor resistance decreased drastically at the moment of water dropping, after that switching on and off the airflow to perform airflow detection, sensor resistance failed to recover to the initial state and lost airflow detection property (Movie S4, Supporting Information).As shown in the optical overlap diagrams, the reason lies in the adhesion of water droplets on the CNTF surface after water dropping.The direction of distortion induced by water droplets and airflow overlapped, together resulting in the resistance reduction of the CNTF sensor.Since the adhered water droplets could not detach from the CNTF surface, distortion of CNTF introduced by water droplets could not recover, so that the CNTF airflow sensor failed to restore the initial state during the next round of airflow detection and lost detection ability.When  = 90°, now when 5 μL water was dropping onto the CNTF sensor, resistance increased immediately.Switching on and off the airflow to perform airflow detection again, the resistance variation of the CNTF airflow sensor obviously reduced (Movie S5, Supporting Information).In combination of Figure S10 (Supporting Information), it further reveals that during airflow detection of CNTF airflow sensor, as the dropping volume of water increased, the variation value of sensor resistance gradually decreased, accompanied with lowering of sensitivity.When the volume reached 25 μL, the sensor exhibited zero sensitivity to airflow (Movie S6, Supporting Information).According to the optical overlap diagrams, the reason was the friction force between water droplets and the thin film after water dropping and adhered to CNTF surface, which was on the opposite direction of the force of airflow on the sensor.Therefore, at the water dropping moment, resistance of the CNTF sensor elevated.Since water droplets could not detach from the surface, the changing amount of resistance was reduced during airflow detection, and so was the sensor sensitivity.Figure 4j-l are the R-T curves of the CNTF-PPPS-1 airflow sensor subjected to 5 μL water droplets on the surface at an airflow velocity of 2 m s −1 with  being 15°, 45°and 90°, respectively.According to the measured R-T curves, at the instant of water dropping onto the CNTF-PPPS-1 surface, sensor resistance underwent slight changes immediately and soon recovered to the state before water dropping.When performing airflow detection again, the airflow sensor exhibited unchanged properties.Additionally, the investigation on influence of water droplet volume on CNTF-PPPS-1 airflow sensor (Figures S11-S13 and Movies S7-S10, Supporting Information), demonstrated that sensor performance was unaffected even when subjected to 200 water droplets (5 μL each), which agrees with the conclusion mentioned above.Combining optical overlap diagrams, the reason lies in the fact that when water droplets were dropped onto the CNTF-PPPS-1 surface, they separated from the surface immediately after bouncing without altering the sensing performance of CNTF-PPPS-1 airflow sensor.It also proves the airflow detection ability of CNTF-PPPS-1 airflow sensors in rain and high humidity environments, regardless of the tilting angle of the sensor.Raindrops will not affect the performance of the airflow sensors.At present, little research has been conducted on super-hydrophobic airflow sensors.In different environments, water always coexists with gas, suggesting extensive application prospect of CNTF-PPPS-1 super-hydrophobic airflow sensors.
We further build analytic models and molecular models to explore the mechanism for such remarkable sensitive proper-ties with ultra-low limit of detection (≈0.1 m s −1 ), fast response (≈0.2 s), and rapid recovery (≈0.35 s) based on CNTF-PPPS-1.First, the ultra-thin CNTF-PPPS-1 films feature great flexibility and ductility.Surface coarse structures increase contacting areas with airflow on the one hand, and the addition of PVA enhances gas barrier property of the thin films on the other hand (Figure 5a).Second, the conductive framework comprising of multi-wall carbon nanotube networks possesses good conductivity and post-distortion self-recovery ability.To elaborate the sensing mechanism of the CNTF-PPPS-1 airflow sensor, we carried out further mechanism investigation and found that the sensing ability of the sensor mainly came from CNTF-PPPS-1 thin film distortion induced by airflow impact.Since the surface coarse structures were basically non-electrically conductive, the resistance variation is negligible upon distortion, so that general distortion will not affect the sensing property of the sensor except for CNTF distortion introduced by airflow impact.Overall, the conductive framework of the CNTF-PPPS-1 sensor can be regarded as the equivalent circuit (shown in Figure 5b).Resistance of CNTF-PPPS-1 is composed of contact resistance and bulk resistance.The former originates from the CNTF network structure, while the latter contains the intrinsic resistance of single CNT (R cnt ) and PPPS-1 (Rppps-1) in the CNTF network.Since bulk resistance is insensitive to gas, the variation in the total resistance of the circuit can be attributed to the variation in contact resistance.When blowing gas to the airflow sensor, CNTF films undergo distortion, resulting in increased interior CNT-CNT contacting spots by the bending of the CNTF network, which multiply electron transport pathways and thereby reduce resistance (Figure 5d).At the same time, the multi-layer structure of CNTF leads to reduced layer distance of the bending part (Figure 5c), hence increasing Van de Waals contacting spots, further reducing contact potential barrier, and increasing conductivity. [46]The two corresponding resistances of contacting spots are together defined as contacting R cont .From the mechanism analysis above, conclusion can be drawn that applying on/off airflow onto the CNTF-PPPS-1 airflow sensor leads to distortion of the interior sensitive elements of the sensor, and thereby altering conductivity of the sensor, so that the airflow detection property is achieved.We have also utilized molecular dynamic simulations to show the dynamic process of carbon nanotubes under blow.As Figure 5e clearly shows, we can see that the number of contact parts between neighbor carbon nanotubes (blue parts) is increased under blow.

Breathing Detection
The CNTF-PPPS-1 airflow sensor exhibited remarkable performance of high sensitivity, and rapid response and recovery in airflow detection.On that account, we also explored the sensing properties of the airflow sensor toward breathing, based on the consideration that airflow variation exists in the process of breathing.Breathing is one of the vital characteristics of human life, and a significant pathway for substance exchange between the human body and the environment.Therefore, breathing detection constitutes an important part of human body detection.[49][50] Breathing detection mainly utilizes airflow variations during breathing, namely the processes of breathing in and breathing out.Sensors can be integrated onto face masks to carry out breathing detection through a bluetooth module at any time.In this way, people can learn about their body status through breathing detection at anytime, anywhere.The CNTF-PPPS-1 airflow sensor also exhibited superb performance in breathing detection.As shown in Figure 6a, resistance variation of the sensor increased gradually during the breathing out phase, while decreased gradually during the breathing phase.In one breathing process, the variation amount of sensor resistance showed a regular change over breathing regularity.The major indicators for breathing measurement are respiratory frequency and respiratory depth.The former is measured by respiration pulses in one minute (namely, respiration counts per minute (60/ΔT), unit: bpm).Generally, counts of respiration per minute equals to 1/4 of that of heartbeats.A normal adult breathes 12-20 times per minute.[53] Respiratory depth is defined by the peak to peak amplitude measured by breathing sensors (ΔR), which is used to estimate respiratory capacity (Figure 6,c).As shown in Figure 6b, the R-T curve of sensors responded to normal breathing demonstrates reduced sensor resistance upon inhalation and increased resistance upon exhalation.Through the whole breathing process, sensor resistance R-T curve changed regularly with the respiration process, exhibiting sensitivity and instantaneity of the sensor.Figure 6d is the R-T curve of sensors detecting the fast breathing-deep breathing-normal breathing pattern of the same person.The respiratory frequency and respiratory capacity varied significantly among fast breathing, deep breathing, and normal breathing.The R-T curve of sensors can obviously reveal distinction of respiration, again demonstrating the instantaneity and sensitivity of the sensor.Meanwhile, the unchanged sensor's R-T curve under the same breathing pattern also reflects the stability of the sensor.In Figure 6e, the sensor was used to detect the breathing of three normal individual persons of similar ages.According to the R-T curve of the sensor, despite that the respiratory frequencies three persons all fell within the range of that of normal people, their breathing status showed great variations, especially in respiratory capacity and the regularity of the breathing curves.It is mainly resulted from the breathing behavior and cardio-pulmonary function of individuals.Therefore, the sensor can be applied for breathing recognition.We adopted the CNTF-PPPS-1 airflow sensor to detect res-piratory characteristics mostly based on the airflow generated by respiration.The principal sensitive elements in the CNTF-PPPS-1 airflow sensor are CNTF, which is a network structure composed of CNT.CNTF resistance shows a negative correlation with temperature.Its resistance decreases with the increase of temperature.During exhalation, temperature around the sensor is elevated, thus decreasing the sensor resistance, [54] as is consistent with the resistance reducing of the CNTF-PPPS-1 airflow sensor.Therefore, investigation was conducted to determine that one of airflow and temperature plays the predominant role in the sensing of the CNTF-PPPS-1 airflow sensor during respiratory process.The temperature of exhaled gas is approximate to body temperature, which is ≈36-37 °C.On that account, moving fingers within the range of 0.5-1.5 cm above the sensor was adopted to simulate temperature change, considering similar temperature of fingers and exhaled gas.Notably, the experiment was carried out in a 23 °C thermostatic room.The results are shown in Figure 6f and g.While sensor resistance exhibited regular variations as fingers moving up and down above the sensor, the variation amount was only 1/6 of that during respiratory processes.Therefore, it proves that detection of respiratory characteristics of the CNTF-PPPS-1 airflow sensor was mainly influenced by airflow, whereas temperature added to the sensing properties of the sensor.The CNTF-PPPS-1 airflow sensor exhibited advantages of instantaneity, stability, rapidity, and reliability, which make the sensor capable of breathing recognition.

Conclusions
In this study, a flexible bridge-type super-hydrophobic carbon nanotube fiber thin film airflow sensors was developed, which not only own remarkable sensitive properties with ultra-low limit of detection (≈0.1 m s −1 ), fast response (≈0.2 s), and rapid recovery (≈0.35 s), but also exclude the influence of moisture or water drops in real applications.CNT film was adopted as the fundamental conductive framework, and super-hydrophobic structures were introduced on the surface of CNT film.Through ingenious combination with the bridge-type flexible substrate, sensors for airflow detection were assembled.The sensor exhibited high sensitivity, good stability, excellent reliability, multi-angle detection, wide detection range, and rapid response and recovery ability.It was demonstrated that the sensor was capable of detecting airflows with a rate of 0.1 m s −1 , with a response time of 0.2 s and a recovery time of 0.35 s.Meanwhile, the unique superhydrophobic structure endowed the sensor with capability of airflow detection in rain and high humidity environments.Apart from remarkable performance in airflow detection, the sensor also showed good sensitivity, instantaneity, rapidity, and detection accuracy in breathing detection, making it potential candidate for monitoring human cardio-pulmonary function and sleep quality, as well as breathing recognition, hence showing extensive application prospect of the sensor.

Figure 1 .
Figure 1.Sensitive element preparation, SEM images, EDS results, working principle and performance of flexible bridge-type super-hydrophobic carbon nanotube fiber thin film airflow sensors.a) Preparation process of CNTF-PPPS and model illustration of water droplets on super-hydrophobic CNTF-PPPS surface.b) Physical and c) SEM images of CNTF (EDS result is inserted).d) Physical and e) SEM images of CNTF-PPPS (EDS result is inserted).f) Working principle of CNTF-PPPS airflow sensors.i) Response curve of CNTF-PPPS airflow sensors to on/off airflows.h) response curve of CNTF-PPPS airflow sensors to breathing.
are the optical overlap diagrams of the movement of water droplets on CNTF-PPPS-1 surface with  being 15°, 45°and 90°, respectively (Movie S3, Supporting Information).When dropped onto

Figure 3 .
Figure 3. CNTF-PPPS and CNTF airflow sensor performance.a) Scheme of the airflow detection sensor and device. stands for the angle between airflow direction and the sensing unit thin film.b) Comparison between resistance change (|R-R 0 |)/R 0 (%) and airflow detected by airflow sensors fabricated by CNTF, CNTF-PPPS-0.5,CNTF-PPPS-1, and CNTF-PPPS-2.c) Response time d) Recovery time comparison among CNTF, CNTF-PPPS-0.5,CNTF-PPPS-1, and CNTF-PPPS-2 airflow sensors.e) CNTF-PPPS-1 airflow sensor at the airflow rate of 0.1-4 m s −1 switch gas under the measurement of sensor curve.f) Sensing curves measured by the CNTF-PPPS-1 sensor at different airflow angles.g) Sensing curve measured by the CNTF-PPPS-1 sensor repeatedly at different airflow angles (2 m s −1 of airflow velocity).h) Response time and recovery time of the CNTF-PPPS-1 sensor at a airflow velocity of 2 m s −1(enlarged image of the dashed box in (g)).i) Sensing curve of 600 times of airflow on/off switch cycles measured by the CNTF-PPPS-1 sensor at a airflow velocity of 2 m s −1 ,  = 90°.j) Comparison of limit of detection and response time between the CNTF-PPPS-1 airflow sensor and airflow sensors reported in references.[31,[37][38][39][40]26,[41][42][43][44][45] (Note: Th negative sign in (R-R 0 )/R 0 value represents only the direction of resistance change, while negative sign indicates the decrease of resistance)

Figure 4 .
Figure 4. Investigation on the influence of water droplets on CNTF and CNTF-PPPS-1 airflow sensors.a-c) The optical overlap diagrams of the movement of water droplets on CNTF surface with  being 15°, 45°and 90°, respectively.d-f) The optical overlap diagrams of the movement of water droplets on CNTF-PPPS-1 surface with  being 15°, 45°and 90°, respectively.g-i) The R-T curves of the CNTF airflow sensor subjected to 5 μL water droplets on the surface at an airflow velocity of 2 m s −1 with  being 15°, 45°and 90°, respectively.j-l) the R-T curves of the CNTF airflow sensor subjected to 5 μL water droplets on the surface at an airflow velocity of 2 m s −1 with  being 15°, 45°and 90°, respectively.(Note: the volume of each water droplet was 5 μL. 25 μL means dropping 5 consecutive water droplets.)

Figure 5 .
Figure 5. CNTF-PPPS-1 airflow sensor mechanism analysis model diagram.a) Structure diagram of CNTF-PPPS-1 applied air flow model.b) CNTF-PPPS-1 sensor equivalent circuit model.c) Schematic diagram of CNTF-PPPS-1 stress deformation after blowing.d) CNT-CNT air flow impact forming contact point process and circuit model before and after contact point formation.e) Molecular dynamic simulations to show the dynamic process of carbon nanotubes under blow.The number of contact parts between neighboring carbon nanotubes (blue parts) is clearly increased under blow.

Figure 6 .
Figure 6.Sensing performance of the CNTF-PPPS-1 airflow sensor in breathing detection.a) Variation curve of sensor resistance variation amount with breathing (inserted is the enlarged image of the red dashed box).b) R-T curve of sensors responded to normal breathing.c) Enlarged from the shallow blue shade part in b. d) R-T curve of sensors responded to the fast breathing-deep breathing-normal breathing pattern of the same person (inserted is the enlarged images of the corresponding breathing pattern).e) R-T curve of sensors responsive to the breathing of three individual persons.f) comparison between the R-T curves of sensors responded to breathing and finger moving sensors (note: finger moving refers to moving the sensor with fingers within the distance of 0.5-1.5 cm above the sensor).g) enlarged image of the dashed box in f.