Potassium Iodide-Functionalized Polyaniline Nanothin Film Chemiresistor for Ultrasensitive Ozone Gas Sensing

Polyaniline (PANI) nanostructures have been widely studied for their sensitivity to atmospheric pollutants at ambient conditions. We recently showed an effective way to electropolymerize a PANI nanothin film on prefabricated microelectrodes, and demonstrated its remarkable sensing performance to be comparable to that of a one-dimensional nanostructure, such as PANI nanowires. In this work, we report further progress in the application of the PANI nanothin film chemiresistive sensor for the detection of ozone (O3) by modifying the film with potassium iodide (KI). The KI-PANI sensor exhibited an excellent sensitivity to O3 (8–180 ppb O3 concentration rage) with a limit of detection of 230 ppt O3, and exquisite selectivity against active chemicals such as nitrogen dioxide (NO2) and sulfur dioxide (SO2). The sensing mechanism of the sensor relied on iodometric chemistry of KI and O3, producing triiodide (I3−) that partially doped and increased electrical conductivity of the PANI film. The sensitivity and selectivity of the KI-functionalized PANI film demonstrates the potential use for KI-PANI-based O3 sensing devices in environmental monitoring and occupational safety.


Introduction
Ozone (O 3 ) is an oxidative tri-oxygen compound that exists at ppb-to-sub-ppm levels in the tropospheric atmosphere. Even though it is present at low concentrations, O 3 is classified as a dangerous pollutant because it is poisonous to human health (Occupational Safety and Health Administration (OSHA) permissible exposure limit (PEL) of 100 ppb), harmful to crops, and involved in various atmospheric chemistries. The involvement of O 3 in photochemical smog creates a dynamic equilibrium of O 3 with other toxic pollutants, including nitrogen dioxide (NO 2 ), nitric oxide (NO), and some volatile organic species. Moreover, O 3 is a known greenhouse gas that contributes to global climate change. The serious impacts of O 3 on atmospheric quality, biological/ecological systems, and personal health provokes public awareness regarding the detection and monitoring of atmospheric O 3 .
Various approaches are currently employed to determine atmospheric O 3 , including iodometric chemistry [1], spectrophotometry [2], and mass sensitive resonance [3]. These methods require experienced operators, elaborate and expensive lab equipment, and are not suitable for field

Materials and Methods
All chemicals were reagent grade, and used as received. Electropolymerization of PANI nanothin film was performed with standard three-electrode electrochemical cell via a CHI760C electrochemistry work station. The prefabricated gold microelectrode was used as substrate for the electropolymerization.

Sensor Fabrication
The microelectrodes were fabricated on highly-doped silicon substrate as pairs of 100 µm × 200 µm gold contacts, separated by 3 µm gap channel using photolithography. Prior to use, the microelectrodes were cleaned in a 3:7 (v:v) solution of hydrogen peroxide and sulfuric acid, rinsed with DI water, and blown dry under a nitrogen stream. The cleaned substrate with microelectrodes was further heated at 100 • C for 1 h and immediately submerged in a solution of 2% (v/v) octadecyltrichlorosilane (OTS) in toluene (anhydrous) to deposit an OTS self-assembled monolayer. The OTS-silanized silicon dioxide surface became highly hydrophobic after the silanization treatment, and attracted aniline monomer to accumulate on. The hydrophobic silicon dioxide surface helped to promote lateral film growth during electropolymerization [10,11], and was a key to realizing a nanothin film structure.
The microelectrodes acting as working electrode, Ag/AgCl reference electrode, and platinum mesh counter electrode were employed in the electrochemical cell in an electrolyte of 0.5 M aniline and 1 M perchloric acid (HClO 4 ). Constant potentials of 0.6 and 0.8 V were applied individually to each pair of gold electrodes, and were terminated as soon as the chronocurrent curves (electrochemical current versus time) bifurcated from one another. The bifurcation indicated a complete film connection Polymers 2017, 9, 80 3 of 8 across the electrode channel as a result of ohmic current flow. The PANI nanothin film was 10-20 nm in thickness, with a highly porous surface structure [10]. More details on the film surface morphology and electrical characteristics of the nanothin film were previously shown by our group [10]. The film device was bathed in a 5 mM KI (pH~7) aqueous solution for 15 min, during which the KI adsorbed and incorporated in the polymer matrix. The KI-functionalized PANI nanothin film was dried overnight in a low-vacuum desiccant for a future use.

Gas Sensing
Gas sensing experiments were performed inside a temperature-controlled gas sensing apparatus. All the sensor devices were sealed inside a 13 cm 3 glass dome, and were poised at constant potential of 0.5 V. Electrical resistances of the sensor were acquired (chemiresistive mode) via a multichannel field point system, and were recorded as a function of operating time. All mass flow controlled streams were combined and mixed together in a tubular static mixer and delivered at a total flow rate of 100 cm 3 /min. Gas sensing parameters, including gas flow rate, time of analyte exposure, and time of the response recovery were assigned via the LABVIEW ® program (National Instruments, Austin, TX, USA), installed on a personal computer. O 3 gas was generated from dry synthetic air (99% purity, Airgas Inc., Long Beach, CA, USA) using an OREC ozone generator (OSMONICS, SP series). The amount of O 3 produced from the generator was pre-calibrated via iodometric procedures [1]. The generated O 3 stream was diluted in synthetic air and immediately introduced to the sensor devices. The NO 2 and SO 2 were introduced from two separate certified gas cylinders (Airgas, Long Beach, CA, USA, 10 ppm).
All PANI film sensors were stabilized under continuous flow of air at a desired temperature and relative humidity before being exposed to analyte pulses. Each pulse included 15 min of analyte exposure followed by 35 min of analyte-free purging period. The windows of analyte concentration were fixed judiciously in this experiment considering suggested values in the national ambient air quality standard (NAAQS) that were regulated by the USA environmental protection agency (EPA). The values were 75, 100, and 75 ppb for O 3 , NO 2 , and SO 2 , respectively.

Sensing with Pristine PANI Nanothin Film
As shown in our previous works [9 -11], electropolymerization of PANI, performed by poising the two gold electrodes at 0.6 and 0.8 V vs. Ag/AgCl, produced a uniform 10-20 nm-thick carpet-like morphology PANI film of high surface area bridging the 3-µm gap channel between the gold electrodes (figure not shown).
Next, we evaluated the sensing performance of the pristine PANI sensor to gas pulses of NO 2 , SO 2 , or O 3 . The sensing results are presented as a calibration plot of normalized resistance change (∆R/R 0 (%)) versus analyte concentration ( Figure 1). As illustrated in the figure, pristine PANI film-based sensors yielded no significant response to SO 2 , which is a weak reducing agent. On the other hand, strong oxidizing agents O 3 and NO 2 produced a significant increase in electrical resistance of the sensors, attributed to degradation of the PANI polymeric chain [13,14]. Further, while the sensing response from the pristine PANI nanothin film to O 3 gas was strong, there was a significant cross reactivity from NO 2 . This interference would indeed be a problem in the detection of atmospheric ozone.

Sensing with KI-PANI Nanothin Film
To alleviate NO2 interference, we investigated the incorporation of iodometry in PANIchemiresistor. The PANI film was modified with KI and employed for gas sensing at 25 °C and 50% relative humidity (RH). As shown in Figure 2a, KI-PANI devices exhibited a remarkable reduction in electrical resistance upon exposure to 7-180 ppb of O3 gas pulses. The KI-PANI sensor showed a wide concentration range that was linear up to 35 ppb (blue trace), and a linear regression equation of Y = −0.566X + 2 was obtained ( Figure 2b). The sensitivity-slope of the regression equation-of ~0.566 %/ppb O3 surpassed the sensitivity of sensors based on metal oxide (~1.50 × 10 −4 %/ppb O3) [6] and single-walled carbon nanotubes (~0.23 %/ppb O3) [8]. In addition, the recovery of sensing response was very good during the analyte-free purging period after exposure to less than 40 ppb O3. An incomplete recovery at higher O3 concentrations may be ascribed to permanent oxidation of PANI. The limit of detection (LOD)-the lowest concentration of the substance that the device can distinguish from a blank sample with 99% confidence (calculated by LOD = 3 SD/s, where SD is the standard deviation of the noise and s is the slope of the linear part of the calibration curve)-was determined to be 230 ppt of O3.

Sensing with KI-PANI Nanothin Film
To alleviate NO 2 interference, we investigated the incorporation of iodometry in PANI-chemiresistor. The PANI film was modified with KI and employed for gas sensing at 25 • C and 50% relative humidity (RH). As shown in Figure 2a, KI-PANI devices exhibited a remarkable reduction in electrical resistance upon exposure to 7-180 ppb of O 3 gas pulses. The KI-PANI sensor showed a wide concentration range that was linear up to 35 ppb (blue trace), and a linear regression equation of Y = −0.566X + 2 was obtained ( Figure 2b). The sensitivity-slope of the regression equation-of~0.566%/ppb O 3 surpassed the sensitivity of sensors based on metal oxide (~1.50 × 10 −4 %/ppb O 3 ) [6] and single-walled carbon nanotubes (~0.23%/ppb O 3 ) [8]. In addition, the recovery of sensing response was very good during the analyte-free purging period after exposure to less than 40 ppb O 3 . An incomplete recovery at higher O 3 concentrations may be ascribed to permanent oxidation of PANI. The limit of detection (LOD)-the lowest concentration of the substance that the device can distinguish from a blank sample with 99% confidence (calculated by LOD = 3 SD/s, where SD is the standard deviation of the noise and s is the slope of the linear part of the calibration curve)-was determined to be 230 ppt of O 3 .

Sensing with KI-PANI Nanothin Film
To alleviate NO2 interference, we investigated the incorporation of iodometry in PANIchemiresistor. The PANI film was modified with KI and employed for gas sensing at 25 °C and 50% relative humidity (RH). As shown in Figure 2a, KI-PANI devices exhibited a remarkable reduction in electrical resistance upon exposure to 7-180 ppb of O3 gas pulses. The KI-PANI sensor showed a wide concentration range that was linear up to 35 ppb (blue trace), and a linear regression equation of Y = −0.566X + 2 was obtained (Figure 2b). The sensitivity-slope of the regression equation-of ~0.566 %/ppb O3 surpassed the sensitivity of sensors based on metal oxide (~1.50 × 10 −4 %/ppb O3) [6] and single-walled carbon nanotubes (~0.23 %/ppb O3) [8]. In addition, the recovery of sensing response was very good during the analyte-free purging period after exposure to less than 40 ppb O3. An incomplete recovery at higher O3 concentrations may be ascribed to permanent oxidation of PANI. The limit of detection (LOD)-the lowest concentration of the substance that the device can distinguish from a blank sample with 99% confidence (calculated by LOD = 3 SD/s, where SD is the standard deviation of the noise and s is the slope of the linear part of the calibration curve)-was determined to be 230 ppt of O3.   The mechanism of sensing in the KI-PANI film sensor is based on the iodometric chemistry between KI and O 3 . In the presence of water, iodide (I − ) from KI reacts with O 3 , yielding iodine (I 2 ) as a reactive species (Equation (1)). The I 2 then reacts with iodide (I − ), based on the Lewis acid-base reaction, and converts into triiodide I − 3 (Equation (2)). The I − 3 further acts as dopant to the PANI, causing an increase in the film conductivity [14,15], as observed in sensing responses ( Figure 2). However, the formation of I − 3 is an endergonic reaction (∆G > 0) which is reversed in the absence of O 3 due to the equilibrium shift upon ceasing of I 2 production. This leads to a response recovery in the O 3 -free environment. It is worth mentioning that Stergiou and his team [16] reported a pH increment in an unbuffered solution of KI upon exposure to O 3 , as a result of hydroxyl (OH − ) production. This phenomenon was not observed in our experiment due to a dominating effect from I − 3 doping.

Effect of Humidity and Temperature
Humidity is an important factor for the sensor performance, since atmospheric humidity controlled the amount of water content in the PANI film. A sufficient amount of water was needed to avoid self-passivation of KI due to a formation of KIO 3 [17,18]. Water dependency of the KI-O 3 reaction led to an investigation of humidity and temperature effects on the sensing performance.
The KI-PANI devices were fabricated and stabilized under fixed temperature and relative humidity. The results showed that response intensity of the KI-PANI device to O 3 gas decreased as the RH decreased from 50% to 35% and 10% (Figure 3a). At 10% RH, the KI-PANI sensors responded to O 3 pulses in positive/opposite direction; i.e., the resistance increased, revealing a domination of chemical oxidation reaction between PANI film and O 3 over the iodometric reaction. On the other hand, temperature increase caused a reduction in the water content in the PANI film, and lowered the sensing performance of the sensors (Figure 3b) as a result of the self-passivating effect on the KI surface. This could also be attributed to the exothermic nature of the KI-O 3 (acid-base) reaction. The mechanism of sensing in the KI-PANI film sensor is based on the iodometric chemistry between KI and O3. In the presence of water, iodide (I − ) from KI reacts with O3, yielding iodine (I2) as a reactive species (Equation (1)). The I2 then reacts with iodide (I − ), based on the Lewis acid-base reaction, and converts into triiodide ( ) 3 I − (Equation (2)). The 3 I − further acts as dopant to the PANI, causing an increase in the film conductivity [14,15], as observed in sensing responses ( Figure   2). However, the formation of 3 I − is an endergonic reaction (∆G > 0) which is reversed in the absence of O3 due to the equilibrium shift upon ceasing of I2 production. This leads to a response recovery in the O3-free environment. It is worth mentioning that Stergiou and his team [16] reported a pH increment in an unbuffered solution of KI upon exposure to O3, as a result of hydroxyl (OH − ) production. This phenomenon was not observed in our experiment due to a dominating effect from 3 I − doping. -

Effect of Humidity and Temperature
Humidity is an important factor for the sensor performance, since atmospheric humidity controlled the amount of water content in the PANI film. A sufficient amount of water was needed to avoid self-passivation of KI due to a formation of KIO3 [17,18]. Water dependency of the KI-O3 reaction led to an investigation of humidity and temperature effects on the sensing performance.
The KI-PANI devices were fabricated and stabilized under fixed temperature and relative humidity. The results showed that response intensity of the KI-PANI device to O3 gas decreased as the RH decreased from 50% to 35% and 10% (Figure 3a). At 10% RH, the KI-PANI sensors responded to O3 pulses in positive/opposite direction; i.e., the resistance increased, revealing a domination of chemical oxidation reaction between PANI film and O3 over the iodometric reaction. On the other hand, temperature increase caused a reduction in the water content in the PANI film, and lowered the sensing performance of the sensors (Figure 3b) as a result of the self-passivating effect on the KI surface. This could also be attributed to the exothermic nature of the KI-O3 (acid-base) reaction. While the effects of humidity and temperature on the sensor response may seem to be a problem, they can be solved by controlling the air sample humidity by humidifying/dehumidifying and establishing calibration plots at different temperatures for the determination of O3 concentration. While the effects of humidity and temperature on the sensor response may seem to be a problem, they can be solved by controlling the air sample humidity by humidifying/dehumidifying and establishing calibration plots at different temperatures for the determination of O 3 concentration.

Cross-Reactivity
Since the KI-PANI sensor device relies on iodometric chemistry, it may be susceptible to interference from reactive gases such as NO 2 and SO 2 . To investigate this, the response of the KI-PANI device to NO 2 and SO 2 pulses was studied at 25 • C and 50% RH. A calibration plot (presented in Figure 4) showed no significant response of the KI-PANI device to SO 2 gas, and only a slight response to NO 2 above 25 ppb. This response, however, was opposite to that for O 3 ; i.e., the resistance increased instead of decreased. The sensitivity and LOD were~0.08%/ppb NO 2 and~2.5 ppb NO 2 , respectively. The sensitivity observed in this case was significantly lower than that for O 3 on KI-PANI and also for NO 2 on pristine PANI (Figure 1). A slight increase in the resistance of the device to NO 2 suggested that the oxidation reaction (PANI-NO 2 )-which yielded resistance increase-was more predominant than the iodometric reaction (KI-NO 2 ), which caused a resistance decrease due to partial doping of PANI with I − 3 .

Cross-Reactivity
Since the KI-PANI sensor device relies on iodometric chemistry, it may be susceptible to interference from reactive gases such as NO2 and SO2. To investigate this, the response of the KI-PANI device to NO2 and SO2 pulses was studied at 25 °C and 50% RH. A calibration plot (presented in Figure 4) showed no significant response of the KI-PANI device to SO2 gas, and only a slight response to NO2 above 25 ppb. This response, however, was opposite to that for O3; i.e., the resistance increased instead of decreased. The sensitivity and LOD were ~0.08 %/ppb NO2 and ~2.5 ppb NO2, respectively. The sensitivity observed in this case was significantly lower than that for O3 on KI-PANI and also for NO2 on pristine PANI (Figure 1). A slight increase in the resistance of the device to NO2 suggested that the oxidation reaction (PANI-NO2)-which yielded resistance increase-was more predominant than the iodometric reaction (KI-NO2), which caused a resistance decrease due to partial doping of PANI with -3 I . To assess the interference from NO2 when present together with O3 in an analyte sample, the response of KI-PANI sensor to a mixture of 25 ppb O3 and 50 ppb NO2 was compared to that for 25 ppb O3 only. The result showed that there was an insignificant difference in the response of the sensor between the two samples ( Figure 5). This is attributed to a significantly high thermodynamic preference of KI in reacting with O3 to NO2 gas as evidenced by the very high standard potential, E°, for KI-O3 reaction (E° = 1.409 V) over that of the KI-NO2 reaction (E° = 0.179) [19,20]. As an alternative, to completely eliminate NO2 interference, air could be filtered through a packed bed of potassium permanganate (KMnO4) to eliminate NO2 [7]. In the same manner, manganese dioxide (MnO2) [7] can be used to take out the O3, and provide the O3-free air stream for the sensor purging purposes. Filtering/treating the air/gas sample by KMnO4 and MnO2 would also help to alleviate any baseline drift resulting from the chemical oxidation of PANI. Additionally, the calculated ozone concentration (using the regression equation fitted to the linear region of the calibration plot in Figure 2b) of 23.41 ± 3.89 ppb was in excellent agreement with the actual concentration of 25 ppb used in the experiment; a 93.64% relative accuracy. Further, the excellent reproducible response to successive detection of ozone (last three histograms in Figure 5) demonstrates the excellent stability of the sensors. To assess the interference from NO 2 when present together with O 3 in an analyte sample, the response of KI-PANI sensor to a mixture of 25 ppb O 3 and 50 ppb NO 2 was compared to that for 25 ppb O 3 only. The result showed that there was an insignificant difference in the response of the sensor between the two samples ( Figure 5). This is attributed to a significantly high thermodynamic preference of KI in reacting with O 3 to NO 2 gas as evidenced by the very high standard potential, E • , for KI-O 3 reaction (E • = 1.409 V) over that of the KI-NO 2 reaction (E • = 0.179) [19,20]. As an alternative, to completely eliminate NO 2 interference, air could be filtered through a packed bed of potassium permanganate (KMnO 4 ) to eliminate NO 2 [7]. In the same manner, manganese dioxide (MnO 2 ) [7] can be used to take out the O 3 , and provide the O 3 -free air stream for the sensor purging purposes. Filtering/treating the air/gas sample by KMnO 4 and MnO 2 would also help to alleviate any baseline drift resulting from the chemical oxidation of PANI. Additionally, the calculated ozone concentration (using the regression equation fitted to the linear region of the calibration plot in Figure 2b) of 23.41 ± 3.89 ppb was in excellent agreement with the actual concentration of 25 ppb used in the experiment; a 93.64% relative accuracy. Further, the excellent reproducible response to successive detection of ozone (last three histograms in Figure 5) demonstrates the excellent stability of the sensors.

Conclusions
An O3 sensing device was fabricated here, based on a KI-functionalized PANI nanothin film. The KI-PANI film exhibited a remarkable sensing performance toward 8-180 ppb of O3 analyte, with the LOD value of 230 ppt, surpassing other sensitive elements such as carbon nanotubes [8] and metal oxide [6]. The sensing mechanism relied on the iodometric reaction of KI and O3 in the presence of water that formed iodine, iodide, and triiodide. The triiodide can dope and increase the conductivity of the PANI film. The sensor also showed discrepancy against interferences, including NO2 and SO2.

Conclusions
An O 3 sensing device was fabricated here, based on a KI-functionalized PANI nanothin film. The KI-PANI film exhibited a remarkable sensing performance toward 8-180 ppb of O 3 analyte, with the LOD value of 230 ppt, surpassing other sensitive elements such as carbon nanotubes [8] and metal oxide [6]. The sensing mechanism relied on the iodometric reaction of KI and O 3 in the presence of water that formed iodine, iodide, and triiodide. The triiodide can dope and increase the conductivity of the PANI film. The sensor also showed discrepancy against interferences, including NO 2 and SO 2 .