Ammonia Sensing Performance of Polyaniline-Coated Polyamide 6 Nanofibers.

To understand the properties of polyaniline (PANI), aim gas, and the interaction between them in PANI-based gas sensors and help us to design sensors with better properties, direct calculations with molecular dynamics (MD) simulations were done in this work. Polyamide 6/polyaniline (PA6/PANI) nanofiber ammonia gas sensors were studied as an example here, and the structural, morphological, and ammonia sensing properties (to 50-250 ppm ammonia) of PA6/PANI nanofibers were tested and evaluated by scanning electron microscopy, Fourier transform infrared spectroscopy, and a homemade test system. The PA6/PANI nanofibers were prepared by in situ polymerization of aniline with electrospun PA6 nanofibers as templates and hydrochloric acid (HCl) as a doping agent for PANI, and the sensors show rapid response, ideal selectivity, and acceptable repeatability. Then, complementary molecular dynamics simulations were performed to understand how ammonia molecules interact with HCl-doped PANI chains, thus providing insights into the molecular-level details of the ammonia sensing performances of this system. Results of the radial distribution functions and mean square displacement analysis of the MD simulations were consistent with the dedoping mechanism of the PANI chains.

Introduction

Condun>an class="Chemical">cting polymers have been explored for the applications in sensors recently due to their advantages in easily tunable chemical structures and morphologies compared to inorganic materials. For gas sensors, one of the widely studied aim gases is ammonia gas, a common volatile compound present in the atmosphere and one of the causes for environmental pollutants.[1] It is also a kind of important chemical resource being employed in the cooling systems, production of fertilizer, and the reduction of NO gases in diesel vehicles.[2−4] However, it is also a toxic and flammable gas and a kind of signal of some disease, so it is essential to monitor the concentration level of ammonia in the places where one may release it. Thus, a variety of ammonia sensors based on organic materials have been developed because some general ammonia detectors or sensors always require high work temperature (such as γ-Fe2O3 and γ-Fe2O3–TiO2,[5] WO3,[6] SnO2,[7] etc.[8−11]) or are time-consuming by nature (for example, Fourier transform infrared spectroscopy detectors[12]). Among these organic materials, polyaniline (PANI), a classical conducting polymer[13−15] with low work temperature, good environmental stability,[16] low cost, and high thermal stability, has been demonstrated to be a promising candidate in gas sensors.[17−23] However, PANI is infusible,[24] almost insoluble and non-processable,[25] and its physical and mechanical properties are not satisfactory for some applications. In comparison with other structures, nanostructures, for example, nanofibers, can provide a high specific surface area deriving from their porous nature, which results in them being highly competitive for making novel gas sensors.[26] We reported previously[27,28] PANI-based nanofiber ammonia sensors prepared via coating PANI onto the surface of electrospun nanofibers that act as templates. The obtained PANI-based nanofiber sensors presented ideal ammonia sensing behaviors, working at room temperature and excellent flexibility, attributing to the structure of nanofibers and the sensing properties of PANI in the sensors.

By now, lots of efforts have been done to understand the mechanism and sensing pron>an class="Chemical">cess[29−32] about PANI-based sensors; however, there are rare direct calculations with molecular dynamics simulations about the mechanism and sensing process. Further understanding of the properties of PANI, aim gas, and the interaction between them is essential because it can help us to design sensors with better properties. Computational methods including molecular dynamics (MD) used in materials simulations have made tremendous strides in the last two decades.[33] MD can provide a microscopic insight into the polymers, small molecules, and the interactions between them. Wang et al.[34] investigated the adsorption behaviors of H2O, CO2, CH4, and N2 gases on the calcite surface under reservoir conditions by studying the binding energy, Helmholtz free energy, and radial distribution function (RDF) of the system. The results show that the preferential adsorption order is H2O > CO2 > CH4 > N2. Fatemi and Foroutan[35] studied the freezing behaviors of pure water and a 14% water–salt mixture by MD via the coarse-grained model. The simulation results show that, in the lower temperature than the obtained freezing point, the sodium and chloride ions tend to form network and reject solution, leading to the reduction of water molecule accumulation.

In this work, MD simulations were performed to make a microsn>an class="Chemical">copic insight into PANI-based ammonia sensors, and the polyamide 6/polyaniline (PA6/PANI) nanofiber ammonia sensor was taken as a case study. The PA6/PANI nanofiber ammonia sensor was prepared via electrospinning and in situ polymerization methods. The sensor’s structural, morphological, and ammonia sensing properties (to 50–250 ppm ammonia) were investigated and analyzed. Since the sensing material was HCl-doped PANI in the PA6/PANI ammonia sensors and PA6 nanofibers just offered a nanofibrous structure, only HCl-doped PANI and ammonia gas were introduced into the simulation system in this work. Microscopic structures of simulated HCl-PANI such as porosity and volume were determined, and the mean square displacement (MSD) of ammonia molecules, the RDF of the gas sensing system, and the interaction energy between ammonia molecules and HCl-PANI chains were investigated in MD simulations. The results show that MD simulations provide fairly good compatibility with the experimental data.

Results and Discussion

Surface and Structural Properties of Nanofiber Sensors

As given in Figure , the scanning elen>an class="Chemical">ctron microscopy (SEM) results of PA6 and PA6/PANI nanofibers indicated that the surface of PA6/PANI nanofibers is not as smooth as that of pure PA6 nanofibers, suggesting that the roughness comes from the existence of PANI, which was in situ polymerized on the surface of the PA6 nanofibers. In addition, the diameter of PA6/PANI nanofibers is larger than that of PA6 nanofibers and the PA6/PANI nanofibers maintained an appropriate fiber structure. The large specific surface area of the PA6/PANI fibers would facilitate the exposure to and the diffusion of ammonia vapor.

Figure 1

SEM images of (a) PA6 and (b) PA6/PANI nanofibers.

Figure presents the FTIR spectra of pure PA6 and PA6/PANI nanofibers. The FTIR spectrum of PA6/PANI consists of characteristic bands detected at 1577, 1400, and 1117 cm–1 due to the quinoid ring C=C stretching, benzenoid C–C stretching, and C–N+ stretching vibration modes.[36] The C=O stretching vibration peak of amide is observed at 1637 cm–1 for PA6 and at 1636 cm–1 for PA6/PANI. The peak of the C=O stretching vibration that shifted from 1637 to 1636 cm–1 may be caused by the hydrogen bond and van der Waals forces between PANI and PA6 nanofibers.

Figure 2

FTIR spectra of pure PA6 and PA6/PANI nanofibers.

Surface and Structural Properties of the Nanofiber Models

The model of the HCl-PANI system is given in Figure a, whin>an class="Chemical">ch represents the final trajectory from the MD simulations, and the final box dimensions are 56.4 × 56.4 × 56.4 Å3. The density of HCl-PANI in the final MD simulation run (NPT at room temperature and standard pressure) is given in Figure b, and it indicates that the final density is 1.2 ± 0.002 g/cm3, which is close to the experimental value of about 1.3 g/cm3.[37] In addition to the density, the volume, surface, and pore diameters of the HCl-PANI nanofiber system were calculated in MAPS as well. The free volume is 63,116.7 Å3, or 35.2%, and the free surface is 116,248 Å2. For pore diameters, the largest pore diameter-included sphere is 5.8 Å, the largest free sphere is 1.8 Å, and the largest included sphere along the sphere path is 4.1 Å.

Figure 3

For the HCl-PANI model, (a) the snapshot of the equilibrated HCl-PANI structure (colors: carbon, gray; hydrogen, baby blue; nitrogen, blue; chlorine, yellow) and (b) the density in the final NPT simulations at room temperature and standard pressure.

Gas Sensing Performances of the Sensors

Figure a presents how the resistance of the sensors n>an class="Chemical">changed during one ammonia sensing test. When the ammonia gas was in, the resistance of the sensors increased immediately, and after a moment, it reached a stable value gradually. When the aim gas was out, ammonia gas in the test chamber would be released from the surface of the nanofiber sensors, resulting in the resistance value gradually reaching the initial value. The dynamic response curves in Figure b are similar to those of the changes in resistance, and with the increase in ammonia concentration, from 50, 100, 150, 200 to 250 ppm, the stable response values obtained with the aim gas grew higher. In Figure c, the response values are graphed versus the ammonia concentration, and since the trend has high linearity, we infer that the sensors have ideal stability in the range of 50–250 ppm ammonia. Additionally, the sensors can give fast response to aim gas. The response time (Tres) and comparison results with reported articles are summarized in Table S1.

Figure 4

(a) Change in the resistance during one test, (b) dynamic responses with ammonia concentration varied from 50 to 250 ppm, and (c) response values with the fitting line of the nanofiber sensors.

Selectivity and repn>eatability are important parameters to gas sensors. To investigate the sensor’s repn>eatability in this work, the obtained nanofiber sensors were exposed to 500 ppm ammonia five times. As indicated in Figure a, the final resistances can almost reach the initial ones, thus suggesting that the sensors possess reliable repeatability. In Figure b, it shows the response values to different gases with 250 ppm, and it is obvious that the response to ammonia is much higher than those to acetone, ethanol, and methanol, which can be concluded that the sensors have good selectivity among these gases.

Figure 5

(a) Repeatability and (b) selectivity of PA6/PANI nanofiber sensors.

Figure provides a snapshot of the equilibrated NH3 and HCl-PANI system at room temperature and standard pressure. This figure illustrates the distribution of NH3 molecules relative to HCl and PANI.

Figure 6

Snapshot of the equilibrated NH3 and HCl-PANI system (the bigger ones are NH3 molecules, colors: carbon, gray; hydrogen, baby blue; nitrogen, blue; chlorine, yellow).

The interaction between n>an class="Chemical">ammonia molecules and HCl-PANI chains has a significant effect on the sensing behaviors, so the RDF in each ammonia sensing system was calculated, as illustrated in Figure . The graphs indicate that the structure in the ammonia sensing system displays the long-range disorder. The analysis of the RDF of N atoms in NH3 molecules and Cl, H+, and N+ atoms in the HCl-PANI chains reveals that the preferential adsorption order for the selected atoms is Cl > H+ > N+. Furthermore, the distance order between N atoms in NH3 molecules and the other selected atoms is N–Cl < N–H+ < N–N+, with distances of about 3.1, 4.1, and 5.0 Å, respectively.

Figure 7

Calculated radical distribution functions (RDFs) of the ammonia sensing system.

Because of the interan>an class="Chemical">ctions between ammonia and HCl-PANI chains, the ammonia mobility in these systems is expected to be anisotropic. For a system at equilibrium, the particles will move in accordance with the equations of motion that define the system and, in general, will tend to diffuse away from their original location.[38] As observed in Figure , the slope values of the lines tend to be smaller with the ratio of ammonia to Cl increased, and the specific D values are given in Table .

Figure 8

Mean square displacement of each HCl-PANI system.

Table 1

Diffusion Coefficient (D) and Interaction Energy ± Standard Deviations between NH3 and HCl-PANI (kcal/mol) of each HCl-PANI System

systems with different ratios of NH3 and Cl	D values (cm2/s)	interaction energy between NH3 and HCl-PANI (kcal/mol)
0.33:1.00	6.54905 × e–0.7	–1032 ± 135
0.67:1.00	6.04372 × e–0.7	–1831 ± 442
1.00:1.00	4.59450 × e–0.7	–2307 ± 370
1.33:1.00	3.51649 × e–0.7	–3417 ± 436

The interaction energy repn>resents the nonbonded potential energies (van der Waals and elen>an class="Chemical">ctrostatics) between two entities. The interaction energies between ammonia and HCl-PANI chains in different systems at the end of the MD simulations are also given in Table . With the percentage of ammonia in the system increased, the interaction energy grows significantly. The amount of HCl-PANI is constant, so as more ammonia molecules are added into the system, the interaction increases due to the increase in the amount of ammonia present.

Conclusions

The PA6/n>an class="Chemical">PANI nanofiber sensor fabricated by electrospinning and the in situ polymerization method showed ideal sensing properties to 50–250 ppm ammonia at room temperature with smooth dynamic responses, good selectivity, and acceptable repeatability. The mechanism of ammonia sensing is based on the dedoping mechanism of PANI chains by the NH3 analytes, which was monitored by the increase in the resistance and decreased conductivity. Complementary MD simulation results based on PANI-based ammonia sensors indicated that the density (1.3 g/cm3) of simulated HCl-PANI chains is close to the experimental value, and ammonia molecules in the system preferred to be adsorbed by Cl rather than H+ and N+ in the HCl-PANI chains. The sensing mechanism is consistent with the RDF and MSD analysis of the MD simulations that show a strong interaction between the Cl– dopant and NH3 molecules. This information can be used to direct the design of new sensors with excellent gas sensing properties.

Experimental Section

Preparation of Nanofibers

PA6 (Mw = n>an class="Gene">2.1 × 104 g/mol, Zig Zheng Industrial Co. Ltd. of Taiwan, China) nanofibers were prepared via electrospinning 20 wt % PA6 in a mixture solution of formic acid. The feed rate, high-voltage power supply, and the distance between the metal needle tip and the ground stainless drum, which acted as the collector, were 0.3 mL/h, 20 kV, and 16 cm, respectively. Obtained PA6 nanofibrous membranes were immersed into 230 mL of 1.2 M HCl solution containing a certain amount of aniline for about half an hour to make sure that the aniline molecules can be adsorbed onto the surface of PA6 nanofibers. Secondly, in situ polymerization of aniline was carried out in an ice/water bath at 0–5 °C, beginning by adding 20 mL of 1.2 M HCl solution containing APS slowly. Successive polymerization lasted for 5 h. Furthermore, after that, the prepared PA6/PANI nanofibers were taken out and washed with deionized water and 1.2 M HCl solution five times. At last, the samples were dried at 50 °C for 24 h. All materials used in this study except PA6 were bought from Sinopharm Chemical Reagent Co. Ltd. of Beijing, China.

Ammonia Sensing Tests

Before gas sensing tests, the obtained PA6/PANI nanofibers were pasted onto electrodes and their sensing behaviors were tested with a homemade test system at room temperature, as shown in Figure S1, with a relative humidity of 65 ± 1%. Furthermore, the tested aim gas was ammonia with the concentration varied from 50 to 250 ppm. All gas sensing measurements were conducted under static conditions. The response values were defined as the ratio of (R – R0)/R0, in which R and R0 are the resistances of sensors in testing gas and air, respectively.[39,40] The response time (Tres) was recorded by the time of 90% of maximum response values.[9,41−44]

Computational Details

All modeling works were condun>an class="Chemical">cted with the MAPS software, version 4.0, from Scienomics,[45,46] and the MD simulations were done with a large-scale atomic/molecular massively parallel simulator (LAMMPS) program. The polymer-consistent force field (PCFF) was used.[47] The models were built on a Dell computer with an Intel Core-i7 processor at 2.79 GHz with an 8 GB RAM.

The reason why the PANI-based gas sensor has good selen>an class="Chemical">ctivity of ammonia is that, when PANI-based materials were exposed to ammonia, their conductivity would be changed due to the reversible reaction PANI–H+ + NH3 ⇋ PANI + NH4+. The system was built using a polymer builder and amorphous cell modules in MAPS. We simulated the doping process with the following steps. First, a repeat unit of HCl-PANI, as illustrated in Figure , was constructed. Then, the double bonds between nitrogen and carbon atoms in the diamine units were broken. After that, this repeat unit was saved as a repeat unit template in the system. Using this template, one HCl-PANI chain was built with 30 repeat units. Finally, the bonds between Cl and N were broken, and the charges for the Cl atoms and the corresponding H atoms were manually set to be −1 and +1, respectively.

Figure 9

Chemical structure of the emeraldine base polyaniline and HCl-doped emeraldine salt.

The model of HCl-PANI was generated using the approach of Ostwal and co-authors.[37] A total of 10 polymer chains were used, and the density of HCl-PANI was set to be 0.15 g/cm3 to avoid ring spearings and catenations. The total energy was then minimized with the conjugate gradient method. The density was increased by compressing the simulation box with a high pressure of 0.7 GPa, and MD simulations were carried out using the NPT ensemble (constant number of atoms, pressure, and temperature) for 1 ns. MD simulations were then carried out in the NVT ensemble (constant number of atoms, volume, and temperature) at 1000 K, and after that, the system was cooled down with steps of 50–325 K. At last, the HCl-PANI system was carried out in the NPT ensemble with 1 atm at 300 K for 1 ns.

Ammonia molecules were added into the HCl-PANI system with HCl-PANI molecules frozen, and then, MD simulations were carried out at 1 atm and 300 K to obtain equilibrium of the ammonia molecules. HCl-PANI chains were then released, and MD simulations were carried out with 1 atm and 300 K for 1 ns. The amount of ammonia added into the system depended on the ratios of NH3 and Cl, which were 0.33:1.00, 0.67:1.00, 1.00:1.00, and 1.33:1.00, respectively.

The MSD of the ammonia molecules with respect to their original position was obtained as the second moment of their distribution at t > 0. The MSD and the diffusion coefficient (D) were calculated by the following equation

The MSD was fitted to a line ⟨|r(t) – r(0)|2⟩ = 6Dt + B, where N is the total number of particles, t is the time, and r(t) and r(0) are the positions of particles at t and t0, respectively.