Highly Sensitive and Fast Responding Flexible Force Sensors Using ZnO/ZnMgO Coaxial Nanotubes on Graphene Layers for Breath Sensing

The authors report the fabrication of highly sensitive, rapidly responding flexible force sensors using ZnO/ZnMgO coaxial nanotubes grown on graphene layers and their applications in sleep apnea monitoring. Flexible force sensors are fabricated by forming Schottky contacts to the nanotube array, followed by the mechanical release of the entire structure from the host substrate. The electrical characteristics of ZnO and ZnO/ZnMgO nanotube‐based sensors are thoroughly investigated and compared. Importantly, in force sensor applications, the ZnO/ZnMgO coaxial structure results in significantly higher sensitivity and a faster response time when compared to the bare ZnO nanotube. The origin of the improved performance is thoroughly discussed. Furthermore, wireless breath sensing is demonstrated using the ZnO/ZnMgO pressure sensors with custom electronics, demonstrating the feasibility of the sensor technology for health monitoring and the potential diagnosis of sleep apnea.


Introduction
3][4] Advanced technologies for biomedical signal monitoring can benefit a wide range of applications, from general wellness management to disease treatment, encompassing wearable consumer DOI: 10.1002/adhm.202304140devices to specialized medical equipment.10] It is particularly critical for identifying sleep disorders such as sleep apnea, which nowadays pose an increasing threat to people's health and even burden economics. [11,12]However, the gold standard method for accurate sleep apnea diagnosis in sleep laboratories involves the use of a facemask and associated electronics for respiratory polygraphs. [13]Although effective, these solutions are cumbersome, uncomfortable, and primarily designed for sporadic use in hospital settings.Placing highly sensitive sensors near the nostril or mouth can eliminate the need for bulky equipment, requiring only compact reading electronics.This will greatly benefit most users seeking to monitor their sleep quality and detect possible sleep disorders.
[16][17][18] In a 1D structure, a slight force can induce distortion, creating piezoelectric polarization and thus modulating carrier transport through the nanostructure. [19,20]owever, there are also challenges associated with 1D structures.The high surface-to-volume ratio of this 1D structure makes it prone to surface defects. [21][27][28] Moreover, integration with flexible substrates poses another bottleneck.[31] However, semiconductor nanostructures often require high process temperatures, which are generally incompatible with flexible substrates.
Harnessing multiple types of energy from a single device is highly desirable in energy technologies. [32]The first dual-energy harvesting device, capable of simultaneously harvesting solar and wind energy, was based on zinc oxide (ZnO) nanorods. [33]The wind energy harvesting mechanism relies on the piezoelectric effect, wherein mechanical deformation induced in the nanorods by wind generates a piezopotential.It is a well-established fact that the piezopotential in ZnO nanorods is partially screened by the presence of free charge carriers originating from surface defects and detrimental to the efficient operation of the device. [34]Surface passivation of ZnO proves to be an efficient approach, effectively neutralizing surface-free charge centers. [35]onsequently, this enhances the functionality of the energy harvesting devices based on 1D ZnO nanostructures, ensuring their efficiency.
Here, we report the fabrication of a flexible, highly sensitive force sensor with excellent response speed using ZnO/ZnMgO (zinc magnesium oxide) coaxial nanotubes (NTs) on graphene layers and the demonstration of wireless breath sensing for sleep apnea application.The flexible force sensors were fabricated through the mechanical release of full-inorganic ZnO/ZnMgO/graphene structures with Schottky metal contacts.Employing coaxial ZnO/ZnMgO heterostructure improved both sensitivity and response speed significantly.The mechanism for enhanced performance is discussed in detail based on their junction characteristics.We demonstrate wireless breath sensing using custom reading electronics, highlighting its potential for general sleep apnea monitoring and treatment.

Fabrication of ZnO Nanotubes and ZnO/ZnMgO Core-Shell Heterostructure Pressure Sensor
Highly sensitive, fast response force sensors have been successfully fabricated by using the 1D ZnO/ZnMgO coaxial NTs on graphene layers.For the comparative study, sensors using bare ZnO NTs were also manufactured at the same time.Figure 1a illustrates the complete fabrication process.The chemical vapor deposition (CVD)-grown graphene films were prepared on the SiO 2 /Si substrate.A 50-nm thick SiO 2 layer was deposited using plasma-enhanced chemical vapor deposition (PECVD) which acted as a mask layer to obtain the desired position-controlled growth of 1-D ZnO nanotubes.Electron beam lithography has been used to define hole arrays on the SiO 2 mask, followed by dry and wet etching processes to expose the graphene films selectively.On this template, the position-selective growth of ZnO nanotubes has been performed using the metal organic chemical vapor deposition (MOCVD).As shown in the left panel of Figure 1b, ZnO nanotubes with controlled shape and position were grown on graphene.The measured outer diameters and height of the nanotubes are in the range between 400-600 nm and 11-13 μm, respectively.For the ZnO/ZnMgO coaxial NTs, the ZnMgO conformal shell layer was coated in situ.Scanning electron microscopy (SEM) micrographs of ZnO/ZnMgO coaxial heterostructures (center of Figure 1b) show uniform and conformal coating of nanotubes by the ZnMgO layer.
To investigate the thickness of the ZnMgO shell layer, we investigated the relative change in the lateral size of the ZnO nanotubes before and after ZnMgO coating.Figure S1, Supporting Information shows the SEM micrographs of ZnO nanotubes and ZnO/ZnMgO core-shell heterostructures.From these images, the surface of the coated ZnMgO shell is very smooth and uniform, with an average thickness of about 95 nm.Energydispersive spectroscopy (EDS) elemental mapping has also been performed to acquire information about the constituent elements in the ZnO/ZnMgO core-shell heterostructures.Figure S2, Supporting Information presents the EDS mapping images of the ZnO/ZnMgO nanotube heterostructure.EDS mapping shows a uniform distribution of the elements Zn, Mg, and O, which confirms the successful formation of the ZnMgO shell layer.
To confirm the formation of ZnO/ZnMgO coaxial heterostructures, we conducted low-temperature photoluminescence (PL) spectroscopy.The samples were cooled down to 15 K and excited using the He-Cd laser source with a wavelength of 325 nm. Figure 1d shows the PL spectrum of ZnO nanotubes and ZnO/ZnMgO coaxial nanotube heterostructures within the energy range of 3.2-3.7 eV.From bare ZnO nanotubes, strong nearband-edge emissions were observed at 3.310, 3.361, and 3.365 eV.The observed peaks can be associated with donor bound-exciton PL spectra.On the other hand, from the coaxial nanotube heterostructures, a new emission peak at 3.511 eV was observed along with the emission from the ZnO core.This emission at higher energy is associated with the near band edge emission originating from the ZnMgO shell.By comparing the position of ZnMgO near-band-edge emission to previously published reports, it is inferred that the Mg content within the ZnMgO shell layer is around 20%.These low-temperature PL results confirm the formation of ZnO/ZnMgO heterostructures on graphene.
The structural analysis of ZnO and ZnO/ZnMgO heterostructures has been performed by taking the X-ray diffraction (XRD) spectra of ZnO nanotubes and ZnO/ZnMgO heterostructures.Figure S3, Supporting Information shows the -2 scan of the samples, measured within the 2 angle range of 30-60°.ZnO and ZnO/ZnMgO samples showed strong peaks near 34.25°and 34.4°, respectively, corresponding to (002) planes of the ZnO and ZnMgO crystals.The shift in the XRD peak originated from different lattice constants of ZnMgO and ZnO.From the characterization of the grown nanomaterials such as XRD, Low-temperature photoluminescence, SEM, and EDS, one can conclude that both ZnO and ZnMgO are phase pure and suitable for the fabrication of piezoelectric pressure sensors.
To fabricate the flexible force sensors, the prepared ZnO/ZnMgO nanotubes were spin-coated with polyimide (PI), and the tips were exposed to oxygen plasma treatment.Here, the PI layer acts as a supporting layer to maintain ZnO/ZnMgO/graphene heterostructures and electrically separates the top metal contact from the underlying graphene layers.The right panel of Figure 1b shows the morphology of nanotubes with PI-filled gaps and exposed tips.Figure 1c shows the cross-sectional transmission electron microscopy (TEM) image of the ZnO/ZnMgO nanotubes with the PI coating, where each structure is identified and marked by white arrows.To finalize the fabrication process, a gold (Au) top electrode was deposited on the tip of the nanotubes by thermal evaporation to form Schottky contact, and the PI-coated heterostructures were detached from the host substrate by mechanical peel-off.For the bottom contact, the chromium/Au bilayer was deposited on the backside of the graphene layers.

Schottky Diode Characteristics of ZnO/Au and ZnO/ZnMgO/Au Junctions
We first investigated the characteristics of ZnO/Au and ZnO/ZnMgO/Au Schottky junctions by studying the temperature-dependent current-voltage (I-V) characteristics.Here, the graphene electrode was connected to the ground, and the Au electrode at the tip of the ZnO (or ZnO/ZnMgO) nanotubes was biased, as illustrated in Figure 2a.The measurement was done with varying temperatures from room temperature (25 °C) to 160 °C, and the chamber was evacuated to avoid any unwanted oxidation.Temperature-dependent I-V curves of ZnO/Au and ZnO/ZnMgO/Au junctions are plotted in Figure 2b,c, respectively.In general, higher temperatures lead to higher current levels for both reverse and forward bias regions.Additionally, the ratio of the current at each region, that is, I f /I r , decreased for increased temperatures.Interestingly, the I f /I r value of ZnO/ZnMgO/Au junction was found to be smaller than that of the ZnO/Au junction, indicating poorer rectifying behavior, which is different from what is generally expected.
For the detailed analysis, we further calculated diode parameters and the forward bias characteristics of each I-V curve using the thermionic emission model or so-called Cheung's method.Here, the current of the Schottky diode at the forward region is expressed as, where I is the current, I s is the reverse saturation current, q is the electron charge, k is the Boltzmann's constant, T is the temperature, V is the applied bias, R is the series resistance, and n is the dimensionless parameter or the ideality factor.Here, I s is expressed as, where A eff is the effective area of the diode, A* is the Richardson constant, and ϕ B is the Schottky barrier height of the junction.By plotting the current density J = I/A eff versus d(V)/d(ln J), one can calculate the RA eff and n/ where  = q/kT.Then, ϕ B can be achieved by defining a function of current density, H (J) = RA eff J + nϕ B and plotting H(J) versus J.The overall fitting will determine the key parameters of the diode: n, ϕ B , and R. The detailed method for calculation can be found in Cheung. [36]he ideality factor n, barrier height ϕ B , and series resistance R were calculated as a function of temperature for both ZnO/Au (black dots and lines) and ZnO/ZnMgO/Au (red dots and lines) junctions, as plotted in Figure 2d,f, respectively.For the ZnO/Au junction, the increasing temperature slightly de-creased n from 3.75 to 3.0, increased ϕ B from 0.50 to 0.60 eV, and decreased R from 1.5 to 0.3 kΩ.For the ZnO/ZnMgO/Au junction, the changes and deviations of each parameter were slightly increased, but the overall behavior was similar.As the temperature increases, n spiked up to 4.5, but in general, decreased to ≈3.5.The barrier height ϕ B was ≈4.6 eV at 25 and 40 °C but increased to 0.62 eV.The series resistance R of the ZnO/ZnMgO/Au junction was much higher than the ZnO/Au junction, which was 12 kΩ at RT and decreased to 9.0 kΩ at 160 °C.
The high ideality factor n deviated from unity and the change of n and ϕ B as a function of temperature indicates that the diode characteristics cannot be fully explained with the thermal emis- sion (TE) model.Thus thermionic field emissions (TFE) and field emissions (FE) should be considered. [37]When the TFE or FE model is responsible for the current transport, the forward bias I-V characteristics can be given by, where E 00 is the characteristic energy, which is related to the tunneling probability, and E0 is the tunneling parameter. [38]At small V, the slope of ln(I) versus V is equal to q/E 0 , and thus E 0 is achieved as q/slope.By numerically solving E 0 = E 00 coth(E 00 /kT), one can get E 00 /kT. Figure 2g shows the E 00 /kT as a function of temperature, where the slopes of ln(I) versus V (where V ranged from 0.2 to 0.  2g.From this, we may conclude that TFE or FE is more responsible for the junction characteristics of ZnO/ZnMgO/Au than that of ZnO/Au.For ZnO/Au, the Schot-tky barrier is formed at the ZnO/Au interface, and electrons in the ZnO nanotube are transferred to Au by either TE (red arrow) or TFE, FE (green arrow) processes, as illustrated in Figure 2h.For the ZnO/ZnMgO/Au case, we expect that electrons are mostly located in the ZnO channel due to the higher conductivity of ZnO compared to ZnMgO. [39,40]Therefore, there is more contribution from TFE or FE, while TE is prohibited due to the energy barrier formed by ZnMgO, as illustrated in Figure 2i.

Sensitivity and Response Speed of ZnO and ZnO/ZnMgO Nanotube-Based Sensors
After studying diode characteristics, we thoroughly investigated and compared the force-sensing properties of the fabricated bare ZnO and ZnO/ZnMgO coaxial nanotube-based sensors.We systematically applied a controlled gas flow of inert argon to apply force and avoid any chemical effects.Figure 3a shows the experimental setup used to measure the pressure response of the ZnO/ZnMgO and ZnO pressure sensors.The mass flow controller (MFC) outlet was placed close to the sensors, and MFCcontrolled gas flow was applied.First, I-V curves with and with-out the argon (Ar) gas flow have been recorded to confirm the piezoelectric response from devices.Here, the bias was applied from −3 to 3 V.From Figure 3b,c, it is evident that applying the Ar gas flow results in a decrease in the current.On the other hand, a much more current decrease was observed from ZnO/ZnMgO than ZnO, which can be confirmed from the insets of Figure 3b,c.We conducted cyclic current-time (I-t) response measurements with increasing gas flow to characterize the sensitivity as a function of applied gas flow.Figure 3d,e shows the transient force response of ZnO and ZnO/ZnMgO pressure sensors recorded at a constant voltage bias of 2 V. Here, the current change was calculated as a ratio to the total current without stimulus, that is, Current change ratio = ΔI/I 0 (ΔI: current change caused by gas flow, I 0 : the base current without gas flow.)During the measurement, the flow rate of Ar has been increased from 100 to 500 sccm (standard cubic centimeters per minute) with a step of 50 sccm.In both ZnO and ZnO/ZnMgO sensors, the current change was increasing monotonically as the applied gas flow increased.However, a much more significant current change of 18.1% was observed from ZnO/ZnMgO sensor compared to the 2.1% of ZnO sensor, at the Ar flow rate of 500 sccm.The plot of current change versus Ar flow rate in Figure 3f clearly shows the difference between the two sensors.The overall sensitivities were calculated to be 0.0042%/sccm for ZnO and 0.0362%/sccm for ZnMgO, with linearity errors of 32.4% and 28.5% for ZnO and ZnMgO, respectively. [41]The saturation of sensitivity at stronger stimulus can also be found in similar types of sensors.
Additionally, we investigated the mechanical force response of the ZnO/ZnMgO pressure sensor by applying force using a voice coil motor, as illustrated in Figure S4, Supporting Information.The sensor was subjected to a constant periodic force of 100 mN, and the corresponding force responses were recorded.Similar to observations with force exerted by using Ar gas, the sensor exhibited stable and reproducible pressure responses.Furthermore, the force response of the sensor has also been recorded with a time-varying force.In this case, the impact force ranged from 200 to 500 mN, with a 100 mN force difference in each cycle.As observed before, in the case of force exerted by using the Ar gas, an increase in the impact force resulted in an elevated device response, demonstrating an overall sensitivity of 0.058% per mN.
We further investigated and compared the force response time of the ZnO nanotube sensor and ZnO/ZnMgO core-shell sensor from the transient force response characteristics, as shown in Figure 3g,h.The uniform argon gas pulses (100 sccm) were applied sequentially for a specific time (10 s), and the change of the currents was monitored as a function of time.From the measured I-t curves, we calculated the rise and fall time of ZnO pressure sensor and ZnO/ZnMgO core-shell sensor.Here, the rise and fall time was defined to be the time required to reach 90% of the total current change for a given pressure/release process.From the ZnO sensor, the rise and fall time was measured to be 4.0 and 1.3 s, respectively.In contrast, highly improved rise and fall times of 68 and 99 ms were measured from the ZnO/ZnMgO pressure sensors, respectively, which are 13-58 times faster than the bare ZnO NT sensor.The result indicates that the ZnO/ZnMgO structure gives a much faster response and recovery, which is suitable for recording precise waveforms of external stimuli such as breath, pulse, etc.
We also compared the sensitivity and response time of the previously published reports on breath sensors and, presented in Table 1.It can be confirmed that our ZnO/ZnMgO pressure sensor has a comparatively faster response time, which is essential for the efficient diagnosis of breath signals and related problems.

Improved Response Time of ZnO/ZnMgO Heterostructures and Advantages of the Core-Shell Structure for Force Sensing Applications
The significant enhancement in sensitivity and response speed of the ZnO/ZnMgO NT sensor can be attributed to the combinational effect of improved piezoelectric performance and effective surface passivation, as illustrated in Figure 4. Regarding the higher sensitivity, it has been reported that the piezoelectric coefficient of ZnO increases with the Mg doping within the crystal. [47]The higher piezoelectric constant will result in a stronger piezoelectric field, boosting the current change.However, this alone does not fully explain the significant increase in sensitivity.It is believed that surface passivation of ZnO with ZnMgO serves to separate the channel and piezoelectric charges, thereby reducing the screening effect on piezoelectric charges.In the ZnO nanotube sensor, the thin hollow ZnO nanotubes with a high surface-to-volume ratio serve as electron channels and piezoelectric crystals.Therefore, the structure is prone to a natural screening effect since the nanotube's electrons will screen the piezoelectric charges in the same nanotube.When ZnMgO coating is applied to the nanotube, the channel is mainly formed in ZnO/ZnMgO interface, while the piezoelectric charge is formed at ZnMgO/Au interface.This separation will significantly reduce the screening of piezoelectric charge, hence boosting the force response.The passivation of surface states also explains the rapid response of the ZnO/ZnMgO sensor, which is faster than ZnO.The surface states are distributed over the entire channel in the bare ZnO nanotube sensor.The capture or release process of electrons in these sites during the rise and fall of the currents will lead to delayed responses.On the other hand, in the ZnO/ZnMgO core-shell nanotubes, the channels are passivated.Hence, much fewer electron capture/release processes lead to the immediate response to external stimuli.

Flexibility and Pressure Response of ZnO/ZnMgO Pressure Sensor in Different Bending Conditions
Figure 5 demonstrates the flexibility of the ZnO/ZnMgO pressure sensor, a crucial consideration for its application in wearable breath sensors.Figure 5a-c shows free-standing polyimide-supported ZnO/ZnMgO-graphene heterostructures in flat, bent, and wrapped conditions.The inherent flexibility of the free-standing membrane allows easy deformation into various shapes.The free-standing pressure sensor is transferred onto a flexible supporting substrate for pressure response measurements, as depicted in Figure 5d.The pressure responses are measured in flat and bending conditions to demonstrate flexibility, as illustrated in Figure 5e-g.First, the I-V characteristics of the pressure sensor are measured, indicating no Our workZnO/ZnMgO Breath sensor 99 ms 68 ms 0.0362%/sccm significant differences in the I-V response between flat and bent conditions, as shown in Figure 5h.The pressure response of the sensor is recorded by applying a constant force using Ar gas.
Figure 5i shows no substantial changes in the pressure response when the sensor is bent or in flat conditions.This confirms that the free-standing ZnO/ZnMgO-graphene heterostructures are flexible enough to be transferred onto any flexible substrate without compromising the sensor's pressure response characteristics.The flexibility of the ZnO/ZnMgO pressure sensor allowed us to integrate the sensor easily within the cannula tube for breath recording, which is discussed in detail in the next section.

Breath Sensing Using Cannula Tube Integrated ZnO/ZnMgO Pressure Sensor
Using the ZnO/ZnMgO NT sensor, we demonstrated realtime breath sensing for applications in sleep apnea detection.
To record the breath signals, the pressure sensor has been integrated with the cannula tube, as shown in the left panel of Figure 6a.Electrical connections have been made with the pressure sensor by using copper (Cu) wire, and the device has been inserted inside one of the hoses of the cannula tube, as shown in the photographs in Figure 6a.Thanks to the sensor's small size and flexibility of the graphene-based nanostructures, the device could be inserted into the narrow space inside the cannula tubes without losing its functionalities (center and right panels of Figure 6a).To record the breath response, the cannula hose carrying the sensor has been inserted inside the nostrils, as shown in the right panel of Figure 6a. Figure 6b shows the breath responses (I vs t) of the ZnO/ZnMgO pressure sensor over a long period of time (>1000 s) and zoomed-in plots for short periods of time (100 and 5 s), as shown in the left, center, and right plots, respectively.As shown in the plots of Figure 6b, the current level has been changed under the exhale, and the current flow through the device decreases whenever the breath has been exhaled.When the user intentionally stopped breathing ("stop" marked region), there was no signal except the slow baseline drift (left panel of Figure 6b).Each exhale is captured and easy to identify, as shown in the center panel of Figure 6b.The breath response curve exhibited the same behavior in the case of Ar flow measurement and the mechanical force exerted by the voice coil motor, where the current level decreased when pressure was exerted.This confirms that the pressure response mechanism did not originate from the moisture or temperature effect of the breath.Thanks to the fast response mechanism of ZnO/ZnMgO structure, the sensor captures exhales and short breaks, while  ZnO sensor exhibited delayed responses (see Figure S7, Supporting Information).
To further investigate the possible effect of moisture on the fabricated pressure sensor, we performed experiments to see the impact of different moisture levels on the device.Figure S6, Supporting Information shows the sensor's response to different humidity levels.In this case, no changes in the current have been observed, which confirms that the pressure sensor is inert to humidity and suitable to record breath responses originating from the piezoelectric effect of the ZnO/ZnMgO material system.Also, it is essential to note that ZnO, ZnMgO, and polyimide are biocompatible materials and have minimal health-related concerns.Polyimide is a widely used synthetic polymer material that has been integrated into various kinds of medical devices and implants. [48]The safety of these materials ensures that the fabricated pressure sensor is a suitable candidate for realizing self-diagnosis-based wearable breath sensors.
The stability of the ZnO/ZnMgO pressure sensor has been investigated by heating the sensor at 35 °C for 1 week and measuring the pressure response each day.The intended application of the fabricated pressure sensor is breath monitoring, and the average human breath temperature ranges from 30-32 °C.
Therefore, we choose 35 °C to conduct the stability test of the pressure sensor.Figure S5, Supporting Information presents the pressure response of the sensor undertaken for 1 week each day at 35 °C.The force on the sensor has been applied using the constant Ar gas flow.From this figure, it is evident that there are no drastic changes in the pressure response of the sensor during the 1 week, and the sensor is stable and confirmed its stability.
To demonstrate a wearable device application for sleep apnea monitoring, we developed a wireless breath sensing system comprised of our ZnO/ZnMgO sensor, a Bluetooth-based electronics module, and a Li-ion rechargeable battery, as shown in Figure 6c.The inset of Figure 6c shows the small size of the Bluetooth module.Figure 6d shows how components are configured to provide wireless sensing.Here, the ZnO/ZnMgO sensor and a resistor (R Series ) form a voltage divider.The sensor resistance changes by external mechanical stimuli such as breath will lead to the voltage change between the sensor and R Series .This voltage node is connected to a non-inverting amplifier based on the operational amplifier (OP-AMP), where the amplification is determined as 1 + R F /R i .The output of the amplifying circuit is connected to the integrated analog-digital converter of the Bluetooth-based mod- ule (nRF-52840).This module can transmit the voltage reading to electronic devices such as phones, tablets, or laptops.We used a laptop computer with homemade Python-based software to wirelessly read and record the voltage reading.Figure 6e shows the typical recording sample by this Bluetooth-based system.Here, the researcher mimicked the sleep apnea situation by holding his breath for a while.This stopped breathing is identified as indicated by red markers.

Conclusion
In conclusion, we fabricated highly sensitive, rapid-responding force sensors utilizing position-controlled ZnO/ZnMgO coaxial NTs grown on graphene layers with Au Schottky contacts.The diode properties were investigated by analyzing temperaturedependent I-V characteristics, revealing that more TFE or FE currents were involved in ZnO/ZnMgO devices.Comprehensive characterizations of force sensing performance were conducted by applying Ar gas flow.A comparison study between ZnO and ZnO/ZnMgO devices revealed that ZnO/ZnMgO devices exhibit about ten times higher sensitivity and 13-58 times faster response when compared to ZnO devices.The ZnO/ZnMgO pressure sensor's improved performance was attributed to the ZnO surface passivation by the ZnMgO shell layer.This passivation reduces screening of piezoelectric potential and electron capturing/releasing process.Notably, the fabricated pressure sensor devices were sensitive enough to detect human breath and are suitable for continuous long breath recordings.With customdeveloped wireless electronics, we demonstrated wireless, realtime breath sensing, which promises great potential for monitoring or diagnosing sleep apnea.

Experimental Section
CVD Growth and Transfer of Multilayer Graphene: Multi-layer graphene films were synthesized on Cu foil using CVD method. [49,50]The Cu foil was cleaned in acetone and isopropanol alcohol bath using ultrasonic agitation.The cleaned Cu foils were placed in quartz boat and inserted inside the CVD chamber.The annealing of Cu foil was carried out at 1030 °C for 15 min, with a hydrogen flow rate of 200 sccm while the chamber pressure was maintained at 200 Torr.Following the annealing process, the growth of graphene was carried out by introducing the mixture of hydrogen and methane (CH 4 ), while the chamber temperature and pressure were kept the same.The flow rate of H 2 and CH 4 was set to 100 and 10 sccm respectively.The growth time was 2 h.After the growth, the flow of CH 4 was stopped, and the chamber was cooled down naturally to room temperature while continuing to flow hydrogen into the reactor.
To transfer the graphene onto the SiO 2 substrate, one side of the graphene-Cu foil was protected by spun-coated polymethyl methacrylate (PMMA) layer.The graphene layers on the other side were removed by oxygen plasma treatment.Next, the Cu foil was completely dissolved within the ammonium persulfate aqueous solution, leaving behind PMMAgraphene layers.The PMMA-supported graphene layers were thoroughly cleaned with deionized water (DI water) and transferred onto the SiO 2 substrate.Overnight drying and hot-plate annealing at 200 °C for 4 h were followed.The residual PMMA from the surface of the graphene was removed in the acetone bath for 30 min.
Preparation of Substrates for the Selective Growth of ZnO Nanotubes on Graphene: SiO 2 layer with a thickness of 50 nm was deposited on graphene layers using PECVD to serve as a growth mask layer.To avoid residual growth and improve the selectivity of ZnO on patterned graphene, the SiO 2 mask layer has been annealed at 600 °C in O 2 atmosphere to reduce the defects in as deposited film.Next, the hole arrays having a diameter of 500 nm with a pitch of 4 micron were defined on growth mask layer using the conventional lithography.After the patterning, the exposed area was dry-etched using the tetrafluoromethane (CF 4 ) plasma.The wet etching process in buffered oxide etchant was followed to completely remove the residual oxide layer and reveal the graphene layers.
Growth of ZnO Nanotubes on the CVD Graphene Film: Positioncontrolled ZnO nanotubes have been grown on graphene by using MOCVD.Diethylzinc (DEZn) (purity > 99.995%) and oxygen gas (purity > 99.995%) were used as zinc and oxygen sources respectively.DEZn source bubbler temperature was maintained at −10 °C and carried to the reaction chamber by Ar (purity > 99.999%) during the growth.To prevent a premature reaction, the O 2 gas line was separated from the main gas manifold line.The flow rates of DEZn and oxygen were set at 20 and 40 sccm respectively during the growth.The growth of ZnO nanotubes has been carried out at 670 °C for 90 min and during the growth the reactor pressure was kept at 3.2 Torr.
ZnMgO Coaxial Coating on ZnO Nanotubes: Position-controlled ZnO/ZnMgO coaxial nanotube heterostructure has been obtained on graphene by growing a ZnMgO shell layer on ZnO nanotubes by using MOCVD.During the growth of ZnMgO zinc and oxygen precursors were the same as used for the growth of ZnO nanotubes and also Biscyclopentadienylmagnesium(Cp 2 Mg, purity > 99.995%) has been incorporated as an Mg source.During the growth, the flow rates of DEZn, Cp 2 Mg, and oxygen were set to 9, 30, and 80 sccm respectively.The growth temperature was set to 500 °C while the reactor pressure of the chamber was 5 Torr.

Figure 1 .
Figure 1.Fabrication of ZnO/ZnMgO nanotube-based flexible pressure sensor.a) Step-by-step fabrication procedure of pressure sensor.b) SEM micrograph of position-controlled ZnO nanotubes and ZnO/ZnMgO core-shell heterostructures on graphene layers.c) TEM micrograph of ZnO/ZnMgO coaxial heterostructures.d) Low-temperature photoluminescence (PL) of ZnO nanotubes and ZnO/ZnMgO core-shell hetrostructures.e) Digital photograph of free-standing ZnO/ZnMgO coaxial heterostructures on graphene layers.f) Digital photograph of the sensor after deposition of top and bottom electrodes, followed by transfer onto flexible substrate.

Figure 2 .
Figure 2. Schottky diode characteristics of ZnO/Au and ZnO/ZnMgO/Au structures.a) Schematic illustration of the diode connection.b,c) Temperaturedependent I-V characteristics of the devices, b) from the ZnO/Au junction, and c) from the ZnO/ZnMgO/Au junction.d-f) Diode parameters of the ZnO/Au (black) and ZnO/ZnMgO/Au (red) junctions as functions of temperatures obtained from the forward bias characteristics of each device, d) ideality factor, e) barrier height, and f) series resistance.g) Calculated E00/kBT values as functions of the temperature from ZnO/Au (black) and ZnO/ZnMgO/Au (red) junctions.h,i) Schematic drawing of band edge alignments and electron transport mechanism of h) ZnO/Au junction, i) ZnO/ZnMgO/Au junction.

Figure 3 .
Figure 3. Pressure-dependent I-V characteristics of the device.a) Illustration of the measurement setup, b,c) I-V characteristics under different gas flow, from b) bare ZnO NT-based pressure sensor and c) ZnO/ZnMgO coaxial NT-based pressure sensor.d,e) Transient pressure response of d) ZnO and e) ZnO/ZnMgO pressure sensor.f) Plots of current change as a function of Ar flow rate.g,h) Transient force response and zoomed-in plot of the single cycle, from g) ZnO pressure sensor and h) from the ZnO/ZnMgO pressure sensor.

Figure 4 .
Figure 4. Expected mechanism for higher sensitivity and faster response time.Band edge diagram, contact formation, and expected behavior of a) ZnO NT and b) ZnO/ZnMgO NT-based sensors, respectively.

Figure 5 .
Figure 5. Flexibility of ZnO/ZnMgO pressure sensor.a) Free-standing ZnO/ZnMgO pressure sensor.b) Free-standing ZnO/ZnMgO pressure sensor in bent condition.c) Free-standing ZnO/ZnMgO pressure sensor in wrapped state.d) Free-standing ZnO/ZnMgO pressure sensor after transferred on flexible substrate.e) ZnO/ZnMgO pressure sensor in flat state.f) ZnO/ZnMgO pressure sensor in bent state with bending radius of 12 mm.g) ZnO/ZnMgO pressure sensor in bent state with bending radius of 8 mm.h) I-V characteristics of ZnO/ZnMgO pressure sensor under flat and different bending conditions.i) Pressure responses of the sensor under flat and different bending conditions.

Figure 6 .
Figure 6.Applications for wired and wireless breath sensing.a) Left: photograph of cannula tube integrated pressure sensor inserted inside the nostril for breath sensing.Center, right: photographs of the device and cannula tubes before (center) and after (right) the integration.b) Left: periodic breath response measurement; center: magnified section of breath response in (b); right: a few breath curves under the breathing.c) A video footage of Bluetooth-based wireless breath sensing demonstration.Inset: photographs of the developed Bluetooth module, showing its front and back sides.d) Schematic illustration of the electronic configuration of the module.e) Signal recorded by the wireless monitoring device.Here, the breathing was stopped intentionally (marked by Threat) to mimic the sleep apnea situation.

Table 1 .
Comparison of sensitivity and response time of breath sensors.