Freestanding and Flexible β-MnO2@Carbon Sheet for Application as a Highly Sensitive Dimethyl Methylphosphonate Sensor.

Research on wearable sensor systems is mostly conducted on freestanding polymer substrates such as poly(dimethylsiloxane) and poly(ethylene terephthalate). However, the use of these polymers as substrates requires the introduction of transducer materials on their surface, which causes many problems related to the contact with the transducer components. In this study, we propose a freestanding flexible sensor electrode based on a β-MnO2-decorated carbon nanofiber sheet (β-MnO2@CNF) to detect dimethyl methylphosphonate (DMMP) as a nerve agent simulant. To introduce MnO2 on the surface of the substrate, polypyrrole coated on poly(acrylonitrile) (PPy@PAN) was reacted with a MnO2 precursor. Then, phase transfer of PPy@PAN and MnO2 to carbon and β-MnO2, respectively, was induced by heat treatment. The β-MnO2@CNF sheet electrode showed excellent sensitivity toward the target analyte DMMP (down to 0.1 ppb), as well as high selectivity, reversibility, and stability.

Introduction

There is a growing demand of flexible and portable electronic devices in various fields such as electronic skins, smart clothes, and electronic displays.[1−5] In line with this trend, sensors based on flexible substrates are attracting increased attention. Wearable sensing systems have been developed with the advance of nanoscale materials and a variety of related devices.[6−10] However, the applications of flexible sensors has been limited due to the use of rigid materials as substrates. Therefore, the fabrication of flexible, freestanding substrates is highly desirable. In this context, various substrates such as poly(dimethylsiloxane) and poly(ethylene terephthalate) have been widely studied from both scientific and industrial viewpoints.[11−14] However, these materials are not conductive and require the addition of a transducer on the substrate. This generates a contact surface that causes many limitations and problems in the device manufacture, hindering their practical application as flexible sensors. Thus, there is an urgent need for the development of substrates that are simultaneously flexible, freestanding, and conductive.[15−19]

Currently, most wearable sensors are based on the measurement of physical signal changes rather than on changes in the surrounding environment.[20−23] Nevertheless, the detection of harmful environmental factors is important for human well-being.[24−26] Among such factors, toxic and hazardous gases used in warfare have caused numerous victims with high mortality rates. For example, in 1998, during the massacre of Halavza in Iraq, about 5000 people were killed using hazard gas mixtures comprising organophosphorus compounds such as sarin, tabun, soman, and VX as the main ingredients. Owing to their colorless and odorless characteristics, these gases can paralyze the human body without visible changes. Therefore, the development of a detection technique for organophosphorus gas is an important research area. However, research on this regard is limited due to the high toxicity of these gases. To avoid their direct use, dimethyl methylphosphonate (DMMP) is normally employed as a nerve agent simulant because it exhibits a similar structure and sensing mechanism to those of organophosphorus gases.[27−33]

As electrode materials for flexible sensors, carbon nanomaterials are widely used because of their excellent electrical conductivity and mechanical stability. However, their small size hampers their use as sensor electrode substrates. To circumvent this issue, researchers have resorted to approaches such as coating of carbon on a polymer substrate and fabrication of polymer composites.[34−38] For instance, He et al. synthesized self-assembled graphene using poly(diallyl dimethylammonium chloride), which was then used in flexible dopamine microsensors by coating it on a polyolefin substrate.[36] Zu et al. fabricated highly flexible composites composed of a cross-linked polyvinyl polymethylsiloxane aerogel with carbon nanotubes for application as strain and pressure sensor electrodes.[37]

Herein, we report the fabrication of a β-MnO2-decorated carbon nanofiber sheet (β-MnO2@CNF) for the development of a flexible and freestanding DMMP sensor electrode. As a carbon precursor, a polymer sheet composed of polypyrrole-coated polyacrylonitrile (PPy@PAN) nanofibers was formed by electrospinning and vapor deposition polymerization. Then, amorphous MnO2 nanoplates were attached to the surface through the reduction reaction of a MnO2 precursor and the polypyrrole coating layer. Next, amorphous MnO2 and PPy@PAN were converted to β-MnO2 and carbon, respectively, by carbonization. The as-fabricated freestanding β-MnO2@CNF sheet electrode displayed a remarkably low minimum detectable limit (0.1 ppb) for DMMP with high reversibility, selectivity, and stability. Furthermore, the mechanical properties of the β-MnO2@CNF sheet were investigated, finding that its electrical resistance remained virtually unchanged even after being repeatedly bent more than 100 times.

Results and Discussion

Fabrication of β-MnO2@CNF Sheet

Figure shows a schematic illustration of the fabrication process for the β-MnO2-decorated carbon nanofiber sheet (β-MnO2@CNF). First, a polyacrylonitrile (PAN) sheet consisting of nanofibers of ca. 700 nm in diameter was generated by electrospinning of a 13 wt % PAN solution (Figure a). The electrospun PAN sheet was then soaked in deionized water to prevent tangling due to static electricity. The drenched PAN sheet was pressed at ca. 50 kg to reduce the empty space between the PAN nanofibers. Subsequently, the pressed PAN sheet was soaked in an ammonium persulfate (APS) aqueous solution, which was used as an initiator for decorating the surface with polypyrrole (PPy). In detail, the APS-soaked PAN sheet was exposed to a vaporized pyrrole monomer to induce vapor deposition polymerization. A PPy-coated PAN sheet (PPy@PAN) in which each fiber had a thickness of 760 nm was obtained without aggregation (Figure b). Then, the PPy@PAN sheet was immersed into a 0.01 M potassium permanganate (KMnO4) aqueous solution at 60 °C for 30 min with vigorously stirring to generate disk-type amorphous MnO2 on the PPy@PAN surface (MnO2-PPy@PAN) by means of a redox reaction between PPy and KMnO4, in which PPy acts as a reductant of MnO4– to MnO2 on the surface of the PPy layer.[39] At the beginning, the MnO2 nuclei were latched onto the PPy surface as a result of the large surface energy of PPy and the interactions between PPy and MnO2. As the reaction progressed, the size of MnO2 gradually increased on the surface. Eventually, sheet-like amorphous MnO2 with a thickness of ca. 1 μm was formed on the PPy@PAN surface (Figure c). Then, the sheet was subjected to a two-step heat treatment, i.e., stabilization at 270 °C in air and carbonization at 700 °C under argon gas flow. In particular, the stabilization step promotes the phase transfer of PPy and PAN to branched backbone structures and a ring structure, respectively, thereby enhancing the crystallinity of the carbon structure. It should be noted that a higher concentration of PAN solution (13 wt %) than that normally used (10 wt %) was employed to obtain thick PAN fibers with the aim of increasing the mechanical stability of the carbon sheet, since carbon nanofibers obtained from 10 wt % PAN solution are very brittle at the macroscale, and their reaction with KMnO4 causes destruction of the carbon structure by overoxidation. As a result, a flexible and freestanding β-MnO2@CNF sheet consisting of nanofibers of ca. 500 nm in diameter was fabricated (Figures d and S1).

Figure 1

Illustrative diagram of β-MnO2-decorated carbon nanofiber sheet (β-MnO2@CNF).

Figure 2

Field effect scanning electron microscopy (FE-SEM) and transmission electron microscopy (TEM) (inset) images of each fabrication step: (a) electrospun PAN sheet; (b) PPy-coated PAN sheet (PPy@PAN); (c) amorphous MnO2-decorated PPy@PAN (MnO2-PPy@PAN) sheet; and (d) β-MnO2-decorated carbon nanofiber sheet (β-MnO2@CNF). Inset: scale bars indicate 500 nm.

Illustrative diagram of β-MnO2-decorated n class="Chemical">carbon nanofiber sheet (β-MnO2@CNF).

Characterization of the β-MnO2@CNF Sheet

The chemical composition of the β-MnO2@CNF sheet was investigated using X-ray photoelectron spectroscopy (XPS). Figure a displays the corresponding wide-scan XPS spectra over a range of 0–800 eV. In the spectrum of β-MnO2@CNF, Mn 2p, Mn 3s, and Mn 3p peaks are observed, demonstrating the presence of Mn. In contrast, CNF produced by stabilization and carbonization of the electrospun PAN sheet without any treatment did not display Mn-related peaks. Furthermore, the high-resolution XPS spectra of Mn 2p revealed that the major components of Mn are 2p1/2 and 2p3/2 spin–orbit states, which are observed at 644.5 and 653.3 eV, respectively. This confirms that Mn is mainly present in CNF as Mn4+, which further demonstrates the presence of MnO2 (Figure b). In the O 1s spectra, the largest peak, with a binding energy of 529.84 eV, can be attributed to the Mn–O bond, while those having binding energies of 531.20 and 532.18 eV stem from OH– radical, adsorbed oxygen, or carbonyl group, and adsorbed water on the outside, respectively (Figure c). In the C 1s spectra, the strongest peak at 284.60 eV is attributable to C–C bonds in the carbon structure, while those with binding energies of 285.83 and 288.05 eV can be ascribed to oxygen-containing functional groups (Figure d).[40,41]

Figure 3

(a) Fully scanned X-ray photoelectron spectroscopy (XPS) of the pristine CNF sheet (black) and β-MnO2@CNF sheet (red): (b) Mn 2p, (c) O 1s, and (d) C 1s XPS spectra of the β-MnO2@CNF sheet.

(a) Fully scanned X-ray photoelectron spectroscopy (XPS) of the pristine CNF sheet (black) and βn class="Chemical">-MnO2@CNF sheet (red): (b) Mn 2p, (c) O 1s, and (d) C 1s XPS spectra of the β-MnO2@CNF sheet.

To confirm the change in crystallinity of MnO2 after the heat treatment, X-ray diffraction (XRD) spectra were recorded (Figure a). After the redox chemical reaction step, peaks associated with the crystallinity of MnO2 did not appear in the MnO2-PPy@PAN, whereas in the β-MnO2@CNF spectrum, characteristic diffraction peaks at 2θ angles of 34.98, 40.60, and 58.72° corresponding to the (110), (101), and (211) planes of β-MnO2 were observed.[42,43] In addition, Figure b exhibits Raman spectra in the range 200–2500 cm–1 of the CNF sheets without and with MnO2 decoration. The typical β-MnO2 peaks at 650 cm–1 indicate the stretching mode of octahedral MnO6, and two weak peaks at 362 and 302 cm–1 originate from the bending mode of O–Mn–O or Mn3O4 as a minor portion of Mn2O3.[44,45] Energy-dispersive X-ray spectroscopy (EDX) element mapping analysis also confirmed that Mn and N were uniformly distributed along the surface of an individual nanofiber in the sheet structure (Figure ).

Figure 4

(a) X-ray diffraction (XRD) spectra of MnO2-PPy@PAN sheet (black) and β-MnO2@CNF sheet (red). (b) Raman spectra of pristine CNF sheet (black) and β-MnO2@CNF sheet (red).

Figure 5

(a) FE-SEM and energy-dispersive X-ray spectroscopy (EDX) dot mapping of (b) Mn, (c) N, and (d) O atoms of the β-MnO2@CNF sheet.

(a) X-ray diffraction (XRD) spectra of MnO2-PPy@PAn class="Chemical">N sheet (black) and β-MnO2@CNF sheet (red). (b) Raman spectra of pristine CNF sheet (black) and β-MnO2@CNF sheet (red).

(a) FE-SEM and energy-dispersive X-ray spectroscopy (EDX) dot mapping of (b) Mn, (c) N, and (d) O atoms of the βn class="Chemical">-MnO2@CNF sheet.

Sensing Performance of β-MnO2@CNF Sheet-Based Chemical Sensor

The β-MnO2@CNF sheet was found to detect DMMP immediately according to a charge-transfer mechanism (Figure S2). Thus, when the β-MnO2@CNF sheet is exposed to the DMMP, both CNF and MnO2 are involved in the gas detection process by creating a continuous charge-transfer pathway. The DMMP molecule is a strong electron donor that provides electrons when adsorbed onto the carbon surface. This increases the electrical resistance by lowering the hole concentration in the carbon. The change in hole concentration is further enhanced by the chemical adsorption of MnO2 and DMMP. The latter is adsorbed by charge interaction between the electron lone pair of the methoxy O atom and the vacant orbital of an acidic Mn4+ surface site. As a result, electrons flow from the DMMP to CNF, which causes an increase in the electrical resistance. The synergy of these two effects enables operating at room temperature, whereas sensors based only on metal oxides require higher operating temperatures.

Figure S3 displays the current–voltage (I–V) curve of the β-MnO2@n class="Gene">CNF sheet-based sensor electrode. The linear behavior of the I–V curve reveals the high uniformity of the electrode. In addition, the initial resistance of the sheet was calculated to be ca. 133.3 kΩ by the reciprocal of the slope.

Figure a represents the response of electrodes based on a pristine CNF sheet and the β-MnO2@CNF sheet with sequential exposure to DMMP. The β-MnO2@CNF sensor displayed larger response and higher sensitivity to DMMP (down to 0.1 ppb) than the pristine CNF-based sensor (down to 0.1 ppm) due to the synergistic effect of carbon and MnO2 components. Figure S4 shows the resistance changes of the β-MnO2@CNF electrode as a function of log-scale DMMP concentration. On the one hand, the sensor electrodes exhibited nonlinear changes in response to DMMP concentrations as lower as 0.1 ppb. On the other hand, a linear tendency was observed over a wide range of DMMP concentration (from 0.1 to 0.1 ppm). Accordingly, the β-MnO2@CNF-based sensor electrode exhibited reversible and reproducible responses to various concentrations of DMMP, and the sensing ability increased with the analyte concentration. Furthermore, the limit of detection (LOD) of the β-MnO2@CNF sheet electrode was determined assuming that resistance changes with a signal-to-noise (S/N) ratio below 3 cannot be considered negligible. In the range where the normalized resistance change displayed linear behavior with the analyte concentration, the DMMP sensing was evaluated and the linear fit was performed (Figure S5). Then, a fifth polynomial fit was executed before exposure to the DMMP. The LOD obtained was 0.108 ppb, which corresponds to the lowest response concentration in Figure a.[46]

Figure 6

Reversible and reproducible responses are measured at a current value (10–6 A) with the different carbon sheets. (a) Normalized resistance changes upon sequential exposure to various concentrations of DMMP (black: pristine CNF sheet; red: β-MnO2@CNF sheet). (b) Normalized resistance change of β-MnO2@CNF sheet upon sequential periodic exposure to different DMMP concentrations (black: 100 ppb; green: 10 ppb; orange: 1 ppb; red: 0.1 ppb). (c) Sensing performance of β-MnO2@CNF sheet electrode during 60 days. Measurements were obtained in day intervals. (d) Selectivity responses of the β-MnO2@CNF sheet electrode toward 0.1 ppm of DMMP and 100 ppm of other chemicals.

Stability and selectivity are critical factors for real application of sensor electrodes. Figure b presents the normalized resistance response upon periodic exposure to DMMP as a function of various concentrations. The sensor demonstrated similar resistance change behavior in both response and recovery over four cycles. Moreover, Figure c exhibits the result of a long-term stability test based on the β-MnO2@CNF sheet for DMMP sensors. During a 60-day experiment, only a decrease of 12.76% in normalized resistance change was demonstrated. This is because carbon, which is not easily transformed or decomposed by environmental factors such as humidity, light, and temperature, is the main constituent of the β-MnO2@CNF sheet. Figure d displays the selectivity toward various hazardous chemicals at room temperature. These chemicals were selected as materials with functional groups capable of interacting with the sensor similarly to DMMP. The vapors, which were introduced in the same way as the DMMP, showed lower resistance change because the acceptor group (−OCH3) in DMMP has a higher polarity (3.62 D) than the functional group in other chemicals. Since larger partial charges interact more strongly with Mn4+, the β-MnO2@CNF sheet sensor had great selectivity toward DMMP.

The β-MnO2@CNF sheet must maintain good mechanical stability toward various types of deformation to be applicable to wearable sensing electrodes. Therefore, the electrode was subjected to a flexibility test with various bending angles. The bending angle is defined as the angle between the two straight lines formed by each end and the center of the sheet. As shown in Figure a, despite the distortion, the electrical resistance of the electrode maintained a linear shape. In addition, there was no significant influence on the signal change when sensing with 0.1 ppb DMMP (Figure b). A repetitive bending test was conducted to identify the electrical resistance change (Figure S6). Even after 300 repetitions, the electrical resistance increased by only 18.09%. Considering this result, it can be concluded that the flexible and freestanding β-MnO2@CNF sheet can be applied as a DMMP sensor.

Figure 7

(a) Current–voltage curve with various bending angles of the β-MnO2@CNF sheet electrode. (b) Normalized resistance change of the β-MnO2@CNF electrode upon 0.1 ppb DMMP exposures with various bending angles.

(a) Current–voltage curve with various bending angles of the β-MnO2@n class="Gene">CNF sheet electrode. (b) Normalized resistance change of the β-MnO2@CNF electrode upon 0.1 ppb DMMP exposures with various bending angles.

Conclusions

In summary, a freestanding β-MnO2@CNF sheet was fabricated using electrospinning, chemical reduction, and carbonization for its application as a flexible chemiresistive sensor electrode. The chemical reduction between KMnO4 as a MnO2 precursor and a PPy coating layer was used to introduce amorphous MnO2 on the substrate surface, and then the crystallinity was changed to β-MnO2 by heat treatment. The β-MnO2@CNF sheet electrode detected DMMP with high sensitivity (down to 0.1 ppb) and long-term stability (60 days) at room temperature. Furthermore, the flexible and freestanding properties of the sheet prevented performance degradation upon bending the sensor electrode with various angles. To the best of our knowledge, this study is the first demonstration of a freestanding carbon-based sheet electrode to realize a flexible chemical sensor for a hazardous chemical. Furthermore, this approach can be expected to serve as a platform for the simultaneous sensing of various chemicals with low operating temperature and flexibility by introducing the appropriate components using carbon-based sheet substrates.

Experimental Section

Materials

PAN (Mw = 150 000), n class="Chemical">pyrrole (98%), APS, KMnO4, acetone, benzene, chloroform, ethanol, hexane, toluene, and DMMP were purchased from Aldrich Chemical Co. N,N-Dimethyl formamide (DMF) was purchased from Junsei Chemical Co. Ltd.

Fabrication of the β-MnO2@CNF Sheet

PAN was dissolved in DMF at 70 °C for 4 h, and PAN nanofibers were electrospun from a 13 wt % PAN/DMF solution. During electrospinning, the solution was injected through a stainless steel needle connected to a high-voltage power supply (Nano NC, 10 kV) at 5 μL/min. The electrospun PAN sheet was soaked in APS 3% aqueous solution. The APS-soaked PAN sheet was then exposed to pyrrole monomer vapor at 60 °C and rinsed several times to form PPy@PAN. PPy@PAN was then immersed in 0.01 M KMnO4 aqueous solution and heated for 30 min. After washing several times, the PPy@PAN treated with KMnO4 was stabilized at 270 °C in air and carbonized at 700 °C under Ar gas flow to obtain the β-MnO2@CNF sheet.

Electrical Measurement of β-MnO2@CNF Sheet Gas Sensor

The gas sensor electrode was fabricated by painting silver paste onto both edges of the β-MnO2@CNF sheet. The electrode was set in a chamber and connected to a source meter to monitor the electrical change. In detail, only prepared solutions of DMMP and other chemicals were used for gas generation in all experiments. The organic chemicals were dissolved in DI water, and appropriate concentrations were obtained by successive 1:10 dilutions. For evaporation from the liquid phase, the nitrogen gas stream was passed through a bubbler in a vessel containing an appropriate organic chemical solution and then flowed into the gas chamber (flow rate: 150 mL min–1). The concentration of gases (CG) was regulated by adjusting the odorant concentration in the liquid phase (CL). The evaporated chemicals were prepared by bubbling nitrogen through solutions of different concentrations at room temperature. A three-way valve allowed the flow of gas to be switched from the pure nitrogen source to the evaporated chemical at the same pressure and flow rate. The evaporated chemicals were precisely controlled using a flow system with a mass flow controller (MFC, SEC. 4400 from KNH), pressure controller, and measuring gas chamber. In this work, two lines of MFCs were employed: one for DMMP vapor and the other for N2. The sensing electrode was exposed to DMMP and other gases through a gas tubing source held at a fixed distance of 1 cm above the sensor platform, with a flow rate of 150 mL min–1. The sensor performance was calculated by measuring the normalized electrical resistance change upon applying a constant current of 10–6 A to the electrode. The normalized resistance change was defined by the following equationwhere R and R0 are the real-time resistance and the initial resistance, respectively.

LOD of the electrode was defined as followswhere Y is the selected data point at the baseline of the response versus time curve before exposure of the DMMP (in this study, n class="Chemical">N = 11 data points) and Y is the corresponding value calculated from the fitted curve.

After the sensor electrode was exposed to the hazardous chemicals for sufficient time, the chemicals were purged with an inert gas to recover the conductance. In addition, response time and recovery time were defined as the time required for a sensor to reach 90% of its maximum sensitivity after feeding the chemicals and the time required for a sensor to reach 10% of the preceding sensitivity after purging with the inert gas, respectively.

Characterization

Field emission scanning electron micrographs (FE-SEM) and EDX spectra were obtained using a JEOL 6700 instrument. A JEOL JEM-200CX microscope was used to acquire the transmission electron micrographs. XPS and XRD experiments were performed on JPS-9000MS (JEOL; Mg Kα X-ray source) and M18XHF-SRA (Rigaku SmartLab; λ = 1.5418 Å) instruments, respectively. Raman spectra were acquired on an FRA 1106/S FT-Raman (Bruker) spectrometer and excited with a 514 nm Ar laser. The two-probe method was used to measure the resistance changes according to the number of bending cycles.