non - contact, radio frequency detection of ammonia

Non-contact, radio frequency detection of ammonia with a printed polyaniline sensor N.B. Clark a, * , L.J. Maher b a CSIRO Materials Science and Engineering, Private Bag 10, Clayton South 3169, Victoria, Australia b Detection Systems Pty Ltd., P.O. Box 397, Bayswater 3153, Victoria, Australia article info Article history: Received 9 February 2009 Received in revised form 17 March 2009 Accepted 29 March 2009 Available online xxxx Keywords: Sensor Remote interrogation Radio frequency Polyaniline Inkjet printing Screen printing abstract A novel system for the detection of ammonia was developed by monitoring the conductance of inkjet printed or screen printed polyaniline films with a radio frequency detector. The system has the advantage of non-contact detection of ammonia within sealed packages. Since the sensor is a passive printed film that is externally interrogated, it does not require an internal power source or associated circuitry, and therefore may be a low-cost device suitable for smart packaging applications. When printed on a suitable substrate, the sensor can be cycled several times using heat or a volatile acid to regenerate the polyaniline surface. Crown Copyright Ó 2009 Published by Elsevier Ltd. All rights reserved. 1. Introduction A range of devices are available for the detection of ammonia but most are costly and complex precision instruments. Accord- ingly, there has been continuing interest in simpler, lower-cost designs that could be deployed more widely, such as those based on thin films of polyaniline. However, even these simple polyani- line ammonia sensors generally require internal electrical power and associated circuitry to operate as designed. Polyaniline (PAn) is a polymer that changes conductivity with change in pH as a result of changes in the degree of protonation of the polymer backbone, making it useful as a sensor for volatile bases such as ammonia. Unfortunately, polyaniline in the conduct- ing state is difficult to process because it cannot be dissolved in common solvents or melted below the decomposition tempera- ture. A common approach to overcoming this difficulty is to prepare polyaniline in the form of dispersions suited to conven- tional application methods. One method of preparing aqueous dispersions with high stabil- ity is based on polyaniline nanofibres, as extensively reported by Kaner’s group [1–17]. Information and insights have accumulated over several years but broadly, aqueous dispersions of polyaniline can be formed by manipulating the reaction conditions to favour homogenous nucleation, either by interfacial polymerisation, or more recently, by rapid mixing during the polymerisation reaction induction period, then allowing the reaction to proceed to comple- tion undisturbed. Both methods result in the formation of polyan- iline nanofibres with diameters in the 30–50 nm range, depending upon the acid used in the reaction to protonate the polymer and make it hydrophilic. The polyaniline nanofibres are small enough to be stabilised by the positive charge they carry in aqueous disper- sion. Such stable aqueous dispersions can be used in inkjet ink, but concentrations are often lower than desirable for printing, as several ink layers may be required before useful conductivity is obtained. An alternative approach is the preparation of dispersions with a bulky hydrocarbon component to increase solubility and plasticity; for example, using dodecylbenzenesulfonic acid (DBSA) as a dop- ant as reported by Killard’s group [18–23]. The DBSA acts as both an anionic dopant and as a surfactant that stabilises the polyaniline dispersion against agglomeration. Because of the improved stabil- ity of the polyaniline dispersion, PAn loading concentrations can be higher, which is better for printing. In this paper, we describe the use of a non-contact, radio-fre- quency detector to externally interrogate the conductance of polyaniline films prepared by inkjet and screen printing polyani- line-nanofibre and polyaniline-DBSA dispersions. Since the sensor is externally interrogated, it does not require an internal power source or associated circuitry, allowing the detection of ammonia inside sealed packages. 1381-5148/$ - see front matter Crown Copyright Ó 2009 Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.reactfunctpolym.2009.03.011 * Corresponding author. Tel.: +61 395452259; fax: +61 395452448. E-mail address: noel.clark@csiro.au (N.B. Clark). Reactive & Functional Polymers xxx (2009) xxx–xxx Contents lists available at ScienceDirect Reactive & Functional Polymer s journal homepage: www.elsevier.com/locate/react ARTICLE IN PRESS Please cite this article in press as: N.B. Clark, L.J. Maher, , React. Funct. Polym. (2009), doi:10.1016/j.reactfunctpolym.2009.03.011 2. Experimental 2.1. PAn dispersion preparation DBSA (90%, technical grade, Fluka, Germany), ammonium per- oxydisulfate (Analar, BDH, UK), Teric BL8 surfactant (C12 ethoxy- lated fatty acid alcohol, Huntsman, Australia) and sodium dodecylsulfate (Analar, BDH, UK) were used as received. Hydro- chloric acid (37%, AR-grade, Merck, Germany) was diluted to 1 M with deionised water. Aniline (AR-grade, Univar, Australia) was distilled under vacuum with vigorous stirring to prevent bumping. A PAn-nanofibre dispersion was prepared using the methods described by Huang and Kaner [11]. The purified aniline (3.2 mmol or 0.3 g) was mixed with 10 mL of 1.0 M HCl acid solution. Ammo- nium peroxydisulfate (0.8 mmol or 0.18 g) was mixed into another 10 mL aliquot of the acid solution. The aniline-acid solution was added to the oxidant and the two solutions rapidly mixed for 30 s and then allowed to react undisturbed overnight. The follow- ing day, the polyaniline was washed with water and centrifuged. After three washings, the supernatant liquor had a pH of 3.2 and was strongly green in colour, indicating the presence of polyaniline particles too small to settle. Before use in printing trials, any remaining particles larger than 1 l m were removed by passing the dispersion through a 55-mm glassfibre filter (Whatman GFA, UK) with vacuum assistance. Teric BL8 was added to adjust surface tension to a level within the operating envelope of the printer. PAn-DBSA dispersions were prepared, based on the method reported by Ngamna et al. [21], but with several minor variations. First, the amount of DBSA used was half of that found by Ngamna et al. to be optimum for maximum PAn concentration (i.e. 0.095 M DBSA, rather than 0.19 M; because of the other modifications em- ployed, the stability of the dispersion at this DBSA addition level was acceptable). Second, the reaction was conducted with mixing only during the reaction induction period and none thereafter, as Kaner’s group had shown that continuous mixing during PAn syn- thesis may increase particle agglomeration. Third, ultrasonication was used to disperse the PAn formed prior to filtration, to break up any large particles and maximise dispersion concentration and stability. The procedure used was as follows: DBSA (1.24 g) was added to 20 mL H 2 O, 0.60 g An, 20 mL 1 M HCl and 0.73 g (NH 4 ) 2 S 2 O 8 , and mixed for 30 s, and then reacted undisturbed for 2.5 h. The dispersion formed was dialysed against 600 mL of 0.05 M sodium dodecylsulfate through a cellulose mem- brane (Sigma–Aldrich D9402) for 48 h with two solution changes, then ultrasonicated with a 750 W Sonics Vibracell 1-cm diameter titanium probe (treatment cycle was 1 s on, 2 s off, 70% of maxi- mum amplitude, 10 min treatment time, equivalent to 3.3 min to- tal ultrasonication time) and passed through a 55-mm glassfibre filter (GFA, Whatman, UK) with vacuum assistance. The viscosity of the dispersions was measured with a Brookfield DV II+ viscometer and surface tension was measured with the rod- pull method [24,25]. The contact angle of water drops applied to each of the substrates was measured with a KSV Instruments Ltd. Cam 200. The UV–Vis spectra of the dispersions were collected with a Perkin–Elmer Lambda 35 spectrophotometer, after dilution to comparable absorbance. PAn film thickness was measured on PVC substrates by taking microtome cross sections and imaging them with a Philips XL 30 Field Emission Scanning Electron Microscope. The two PAn dispersions were cast directly on selected sub- strates, or injected into the cartridge of a Dimatix Materials Printer (model 2811 FUJIFILM Dimatix, USA) for inkjet printing trials. An aliquot of the PAn–DBSA dispersion was also evaporated to a paste for application with a screen printing kit (Riso PG-5, Japan). 2.2. Substrates For the application trials, a range of substrates were employed of commercial and general laboratory origin, the majority being polymer label stock that may be candidates for future commercial sensor deployment. Based on Raman spectra (data not given for commercial reasons) the label stock samples were composed of polyethylene with TiO 2 filler, with an ink receptor layer applied to the surface. Substrate pH was measured with a Hanna Instru- ments model 211 pH meter, on solutions prepared by chopping 0.5 g of each substrate into 2 mm strips and soaking in 10 g of deionised water for 72 h. 2.3. Radio frequency conductance measurements A PCIS-3000 10-95 6536 radio frequency detector (Detection Systems, Melbourne, Australia) [26] was used to monitor the con- ductance of sensor strips coated with the polyaniline dispersions. PCIS equipment is typically used to detect missing and damaged items inside sealed packs as they are conveyed on high speed pro- duction lines. This electromagnetic inspection technique is based on a low voltage, low frequency AC electric field. Sealed packages are conveyed through the PCIS transducer by means of a motorised conveyor and a series of ‘slices’ are recorded and assembled into an overall picture of the internal contents of the package. A computer analyses this overall picture for any defects. In traditional PCIS applications it is desirable to focus on the contents and ignore the RF signal generated by the packaging. However, for the exper- iment described here, it was desirable to detect the signal from a sensor embedded in, or on, the packaging. Further, the sensor con- ductivity was modulated by the local environment (i.e. the concen- tration of ammonia inside the box). To our knowledge this is the first time such a device has been used as a component of a RF sen- sor system. 2.4. Lumped circuit approximation An equivalent circuit of the sensor while centred in the scan- ning head can be described as follows (refer to Fig. 1). The trans- mitted signal reaches the receiving electrode by two parallel paths – (1) directly capacitively coupled, and (2) coupled via the sensor. Here, C bypass is the direct coupling from the transmitting Fig. 1. Simple lumped circuit approximation of sensor in the PCIS transducer. 2 N.B. Clark, L.J. Maher /Reactive & Functional Polymers xxx (2009) xxx–xxx ARTICLE IN PRESS Please cite this article in press as: N.B. Clark, L.J. Maher, , React. Funct. Polym. (2009), doi:10.1016/j.reactfunctpolym.2009.03.011 to receiving electrode across the scanning head (substantial rela- tive to the sensor path). C 1 is the coupling from the transmitting electrode to the sensor strip. Z sensor is the impedance of the PAn sensor strip. The sensor strip is mostly resistive in its conducting state. C 2 is the coupling from the top of the sensor strip to the receiving electrode. This model is useful for understanding the ba- sics of the scanning head. Some limitations of this model include:  Impedances involved are very high, circuit elements are distrib- uted and stray capacitances are present.  Packaging materials are neglected. The impedances of materials generally thought of as insulating (paper, cardboard, plastic, or the conveyor belt) can be significant in value, but in practice their effect can generally be mitigated. Their response is gener- ally constant, localised, or at a lower frequency than that of the sensor or product. For the polyaniline sensor tests, a hinged polypropylene box of dimensions 195 mm  55 mm  45 mm was employed, with the sensor strip taped to the inside, vertically near the centre of the longitudinal axis of the box (Fig. 2). The composition of the plastic was not important as the only requirement was for a constant cross section. Three drops of the test liquid, either ammonia solu- tion or acetic acid, were applied by pipette to the inside of the box along the front edge. When closed, the box would quickly fill with vapour from the applied liquid, potentially inducing a change in the degree of protonation of the PAn chains on the sensor surface. Passing the box over the transmitter on the conveyor belt al- lows the conductance of the PAn strip to be monitored, which in turn is an indicator of the atmosphere within the box. The trans- mitter generates a low sinusoidal voltage, typically 8–20 V rms, in the low frequency range. The conveyor was 3-mm thick and was driven by a variable speed drive, but with the same speed used for all experiments. The receiving assembly on the PCIS equipment actually has two ‘lanes’ or inspection zones. Because only one zone was required for these tests the two zones were electrically con- nected together, effectively forming one double sized inspection area. The amplifier converted the received signal to a low imped- ance suitable for subsequent signal processing. The amplified sig- nal was converted to a DC voltage of relatively low frequency. These waveforms are the ‘RF signal vs time’ plots presented in this paper. Positive RF signals resulted if a conductive strip or sensor was placed in the scanning head. The strip or sensor acts as a ‘par- tial reduced impedance path’, compared with the air/bypass path. The computer converted and stored the waveforms to disk. Each waveform consisted of 500 samples over a duration of 2 s (250 Hz sampling rate). The range of the ‘RF signal’ is 0–4095, with a baseline (nothing in the scanning head) around 2270. The stored waveforms were analysed using a spreadsheet. Of particular inter- est was the height of the peak caused by the PAn sensor in the mid- dle of the package. The time series of the height of this peak over time was also of key interest. 3. Results and discussion 3.1. Substrate properties Important properties of the substrates used to support polyan- iline sensor films were the contact angle of applied water drops and pH (Table 1). Contact angle predicts wettability, which in turn may affect the uniformity of the applied polyaniline film, whereas substrate pH may affect the performance of the polyaniline film as an ammonia sensor. Accordingly, the best performance would be expected from substrates that were both hydrophilic (contact an- gles less than 90°) and acidic (pH of <7). Only the glass and filter paper met both these criteria. 3.2. PAn dispersion properties The properties of the prepared PAn dispersions are listed in Ta- ble 2. The PAn-nanofibre dispersion had a total solids concentra- tion less than 1%, and not surprisingly, had properties not dissimilar to water. This meant that surface tension in particular was higher than desirable for the Dimatix printer and two drops of Teric BL8 surfactant (C12 ethoxylated fatty acid alcohol, Hunts- man) was added to reduce surface tension to 37 dyn cm À1 prior to printing. The PAn-DBSA dispersion had a solids concentration al- most two and a half times higher than that of the nanofibre disper- sion, and surface tension was within the range of the Dimatix printer. The UV–Vis spectra were consistent with those expected for PAn-nanofibre [10] and PAn-DBSA dispersions, respectively [21]. 3.3. PAn-nanofibre application Test prints were made with the PAn-nanofibre dispersion on filter paper. Because the PAn concentration was low, print intensity Fig. 2. Test arrangement using PCTS-3000 radio frequency detector. Table 1 Substrates tested. Substrate Supplier Contact angle (°)pH Glass J. Melvin Freed, USA 11 7.0 Filter paper Whatman, UK 28 5.3 PVC Ibico, Australia 87 7.0 Label stock A JAC Australia 118 7.9 Label stock B KW Doggett, Australia 101 7.6 Label stock C Graytex, Australia 119 6.1 Label stock D Avery Dennison, Australia 74 8.1 Label stock E Dalton, Australia 105 8.4 Table 2 Dispersion properties relative to dimatix printer requirements. Liquid Total solids (%) Viscosity (mPa S) Surface tension (dyn/cm) Dimatix ideal n/a 11 30.0 PAn-nanofibre <1 2.3 73.6 PAn-DBSA 2.4 2.8 31.8 N.B. Clark, L.J. Maher / Reactive & Functional Polymers xxx (2009) xxx–xxx 3 ARTICLE IN PRESS Please cite this article in press as: N.B. Clark, L.J. Maher, , React. Funct. Polym. (2009), doi:10.1016/j.reactfunctpolym.2009.03.011 remained poor even after several layers were applied in succession (examples are therefore not reproduced here). When the PAn- nanofibre dispersion was cast (rather than printed) on glass, a thin, uniform film was formed that was soft and easily damaged by handling. 3.4. PAn-nanofibre radio frequency response The PAn-coated glass slide was installed in the test box and passed beneath the radio frequency head several times, with ammonia or acetic acid solutions added successively to change the box atmosphere as described above. The radio frequency wave- form was collected at each pass. Output traces representing acidic and alkaline atmospheres are presented in Fig. 3. Under acidic conditions, a large peak was evident in the output from the radio frequency detector, corresponding to the position of the PAn- nanofibre sensor strip in the box axial dimension. Since the position of the peak relative to the horizontal axis of the chart reflected the physical position of the PAn strip inside the plastic box, moving the sensor strip within the box would result in a cor- responding lateral movement of the peak in the output trace. The PAn peak amplitude varied according to the atmosphere within the box. Peak maxima were obtained either initially, after the sensor strip was first installed with the PAn-nanofibre in the emeraldine salt form, or after exposure to acetic acid following pre- vious exposure to ammonia. Peak minima (i.e. baseline levels) were reached after exposure of the sensor strip to ammonia, caused by de-protonation of the amine groups in the PAn polymer to the emeraldine base form. Overall, the results demonstrated that the PAn-nanofibre sensor worked as reported by Kaner’s group [2], although the low concentration of the PAn-nanofibre dispersion meant that satisfactory prints would require many passes of the print head. Since the detector was sensitive enough to register disturbances in the radio frequency field caused by transitions from air to plastic as the box was conveyed through the detector, two peaks were 0 50 100 150 200 250 Box axis (mm) 2250 2300 2350 2400 2450 2500 2550 2600 2650 RF Signal Acid Alkali Fig. 3. Radio frequency signal obtained from PAn-nanofibre dispersion cast on glass, with successive exposure to ammonia and acetic acid. 2360 2380 2400 2420 2440 2460 2480 0 200 400 600 800 1000 1200 Surface resistivity (KOhms/sq) RF Signal Fig. 4. Relationship between radio frequency signal and surface resistivity measured with a 4-point probe. 4 N.B. Clark, L.J. Maher / Reactive & Functional Polymers xxx (2009) xxx–xxx ARTICLE IN PRESS Please cite this article in press as: N.B. Clark, L.J. Maher, , React. Funct. Polym. (2009), doi:10.1016/j.reactfunctpolym.2009.03.011 evident in the device output corresponding to the two ends of the box, and three smaller bumps on the baseline (at 60 mm, 75 and 90 mm in the axial dimension) associated with shallow decorative mouldings incorporated by the manufacturer in the box design. 3.5. PAn-DBSA application The image intensity obtained by inkjet printing the PAn-DBSA dispersion on filter paper was satisfactory. However, sharpness suffered as a result of lateral migration of the dispersion along the fibres and coverage was not completely uniform, with mottle and print head scan lines both evident. To determine the relation- ship between RF signal and electrical resistance, filter paper was inkjet printed with varying layers of the PAn-DBSA. The surface resistivities of the prints were measured with a 4-point probe and related to the corresponding RF signal (Fig. 4). It was evident that the relationship was linear over much of the surface resistivity range, although at low resistivity the RF Signal appeared to plateau. To achieve a fair comparison, the PAn-DBSA was cast on glass for the initial tests with the radio frequency detector. In this case, the dispersion shrank dramatically on drying, leaving only a small area of the glass covered by the polymer, which appeared to be a film much thicker than that achieved using the PAn-nanofibre dis- persion. Nevertheless, successive exposure to ammonia and acetic acid produced a series of peaks in the radio frequency detector out- put similar to those obtained with the PAn-nanofibre dispersion (Fig. 5). By abstracting the signal maxima of the sensor, it was possible to plot signal strength against time (Fig. 6). The sensor could be cycled by successive exposure to ammonia and acetic acid but on the third such cycle the baseline showed evidence of a rise in con- ductance, probably from accumulation of ammonium acetate. Repeating this process for PAn-DBSA cast on each of the sub- strates listed in Table 1, several observations were made. First, the results were primarily influenced by pH, with those substrates with alkaline surfaces (label stocks A, B, D and E) giving poor 0 40 80 120 160 200 Box axis (mm) 2250 2300 2350 2400 2450 2500 2550 2600 2650 RF Signal Initial signal After ammonia After acetic acid Fig. 5. Radio frequency signal obtained from PAn-DBSA dispersion cast on glass, with successive exposure to ammonia and acetic acid. 2250 2300 2350 2400 2450 2500 2550 2600 2650 0 102030405060708090100 Time (min) RF Signal Fig. 6. Radio frequency maxima obtained from PAn-DBSA dispersion cast on glass, with successive exposure to ammonia (squares) and acetic acid (circles). N.B. Clark, L.J. Maher / Reactive & Functional Polymers xxx (2009) xxx–xxx 5 ARTICLE IN PRESS Please cite this article in press as: N.B. Clark, L.J. Maher, , React. Funct. Polym. (2009), doi:10.1016/j.reactfunctpolym.2009.03.011 results because of partial de-protonation of the PAn-DBSA film be- fore exposure to ammonia. Second, shrinkage of the PAn-DBSA film on drying was a significant issue for smooth substrates like glass and PVC, and also for the non-absorbent label stocks. Overall, the best results were obtained using the filter paper substrate, which was low in pH and had a micro-texture that prevented PAn film shrinkage on drying. Another method of cycling the sensor may be by simply heating the printed substrate in air, driving off the volatile ammonia as re- cently detailed by Crowley et al. [23]. This approach was adopted for PAn-DBSA inkjet printed on filter paper (Fig. 7). Although a sta- ble baseline was obtained because ammonium acetate could not accumulate, the sensor did not recover quickly on heating. The PAn-DBSA dispersion, while more concentrated than the PAn- nanofibre dispersion, was still quite low in viscosity, as required for inkjet printing. A consequence of this low viscosity was deep penetration of the dispersion in the z-direction, almost to the underside of the paper sheet. Such deep penetration could slow the evaporation of ammonia, as the diffusion path through the coated paper matrix would be tortuous. To test this concept, the dispersion was thickened by evaporation, forming a paste that could be applied by screen printing. The thickness of a screen- printed film on PVC was measured from SEM images of cross 2250 2300 2350 2400 2450 2500 2550 2600 2650 0 20406080100120140160 Box axis (mm) RF Signal Fig. 7. Radio frequency maxima obtained from PAn-DBSA dispersion, inkjet printed on filter paper, with exposure to ammonia and recovery by heating in air on a hotplate at 70 °C. Ammonia additions (square points) result in steep declines in the RF signal, followed by gradual recovery with heating. Table 3 Film thickness of PAn dispersions applied to PVC. PAn dispersion Film thickness ( l m) PAn–nanofibre 0.7 PAn-DBSA inkjet ink 4 PAn-DBSA screen print paste 7–25 2250 2300 2350 2400 2450 2500 2550 2600 2650 0 20 40 60 80 100 120 140 Time (min) RF Signal Fig. 8. Radio frequency maxima obtained from PAn-DBSA dispersion, screen printed on filter paper, with repeated application of ammonia (square points) and recovery by heating in air on a hotplate at 70 °C. 6 N.B. Clark, L.J. Maher / Reactive & Functional Polymers xxx (2009) xxx–xxx ARTICLE IN PRESS Please cite this article in press as: N.B. Clark, L.J. Maher, , React. Funct. Polym. (2009), doi:10.1016/j.reactfunctpolym.2009.03.011 sections and compared with inkjet printed PAn-DBSA and PAn- nanofibre (Table 3). Screen printing the PAn-DBSA dispersion onto filter paper gave a rough and thick film compared to other methods of application. However, it gave an ammonia sensor that could be rapidly and repeatedly cycled by heating in air on a hotplate at 70 °C(Fig. 8). 4. Conclusions Ammonia sensors were prepared by printing polyaniline disper- sions on various substrates and using a radio frequency detector to monitor conductance. Dispersions of polyaniline doped with dode- cylbenzenesulfonic acid could be prepared at higher concentration than those of polyaniline nanofibres and consequently the former were more suitable for inkjet printing on a range of substrates. However, these substrates varied in their suitability as sensor sup- ports because of variations in both pH and contact angle, with the best substrate found to be a filter paper with a low pH and a micro- texture that obviated PAn-DBSA film shrinkage on drying. The ink- jet printed PAn-DBSA dispersion worked as an ammonia sensor but cycling time on exposure to volatile acid or heat was slow, proba- bly because of deep penetration of the low viscosity PAn-DBSA ink- jet formulation through the sheet. A screen-print paste prepared by evaporating the PAn-DBSA inkjet formulation to a higher viscosity was more successful, rapidly and repeatedly cycling simply by heating in air. Together, the PAn-DBSA and the radio frequency detector form a non-contact sensor for the detection of ammonia within sealed packages, and is therefore a low-cost device that might be suitable for smart packaging applications. Acknowledgements The authors are indebted to Dr. Orawan Winther-Jensen (nee Ngamna), of Monash University, for advice on the preparation of PAn-DBSA dispersions. The SEM images were collected by Mark Greaves at CSIRO Materials Science and Engineering. Contact angle measurements were made with the assistance of Dr. Christian Kugge of CSME, using equipment located at CSIRO Molecular and Health Technologies. The authors are indebted to Dr. Ken Wong of Scion, New Zealand, Dr. Nafty Vanderhoek and Dr. Warwick Rav- erty at CSIRO Materials Science and Engineering for valuable discussions and suggestions. The work formed part of a New Zea- land Government FRST project entitled Functional Packaging Sys- tems for Food Exports and was financially supported by the Cooperative Research Centre for Functional Communication Sur- faces (CRC SmartPrint) and CSIRO Materials Science and Engineering. References [1] J. Huang, S. Virji, B.H. Weiller, R.B. Kaner, J. Am. Chem. Soc. 125 (2003) 314– 315. [2] J. Huang, S. Virji, B.H. Weiller, R.B. Kaner, Chem. Eur. J. 10 (2004) 1314–1319. [3] J. Huang, R.B. Kaner, J. Am. Chem. Soc. 126 (2004) 851–855. [4] J. Huang, R.B. Kaner, Angew. Chem. Int. Ed. 43 (2004) 5817–5821. [5] H. Gao, T. Jiang, B. Han, Y. Wang, J. Du, Z. Liu, J. Zhang, Polymer 45 (2004) 3017– 3019. [6] D. Li, R.B. Kaner, Chem. Commun. (2005) 3286–3288. [7] S. Virji, R.B. Kaner, B.H. Weiller, J. Phys. Chem. B 110 (2006) 22266–22270. [8] F. Masdarolomoor, P.C. Innis, S. Ashraf, R.B. Kaner, G.G. Wallace, Macromol. Rapid Commun. 27 (2006) 1995–2000. [9] A.Z. Sadek, W. Wlodarski, K. Shin, R.B. Kaner, K. Kalantar-zadeh, Nanotechnology 17 (2006) 4488–4492. [10] D. Li, R.B. Kaner, J. Am. Chem. Soc. 128 (2006) 968–975. [11] J. Huang, R.B. Kaner, Chem. Commun. 1 (2006) 367–376. [12] H.D. Tran, R.B. Kaner, Chem. Commun. (2006) 3915–3917. [13] A.Z. Sadek, W. Wlodarski, K. Kalantar-Zadeh, C. Baker, R.B. Kaner, Sensor. Actuat. A-Phys. 139 (2007) 53–57. [14] A.Z. Sadek, C.O. Baker, D.A. Powell, W. Wlodarski, R.B. Kaner, K. Kalantar-zadeh, IEEE Sens. J. 7 (2007) 2. [15] D. Li, R.B. Kaner, J. Mater. Chem. 17 (2007) 2279–2282. [16] B.J. Gallon, R.W. Kojima, R.B. Kaner, P.L. Diaconescu, Angew. Chem. Int. Ed. 46 (2007) 7251–7254. [17] C.O. Baker, B. Shedd, P.C. Innis, P.G. Whitten, G.M. Spinks, G.G. Wallace, R.B. Kaner, Adv. Mater. 20 (2008) 155–158. [18] S J. Su, N. Kuramoto, Synth. Met. 108 (2000) 121. [19] M.G. Han, S.K. Cho, S.G. Oh, S.S. Im, Synth. Met. 126 (2002) 3. [20] L. Terlemezyan, M. Radenkov, P. Mokreva, P. Radenkov, A.J. Atanasov, Polym. Mater. 20 (2003) 213. [21] O. Ngamna, A. Morrin, A.J. Killard, S.E. Moulton, M.R. Smyth, G.G. Wallace, Langmuir 23 (2007) 8569–8574. [22] K. Crowley, E. O’Malley, A. Morrin, M.R. Smyth, A.J. Killard, Analyst 133 (2008) 391–399. [23] K. Crowley, A. Morrin, A. Hernandez, E. O’Malley, P.G. Whitten, G.G. Wallace, M.R. Smyth, A.J. Killard, Talanta 77 (2008) 710–717. [24] S.D. Christian, A.R. Slagle, E.E. Tucker, J.F. Scamehorn, Langmuir 14 (1998) 3126. [25] S.D. Christian, A.R. Slagle, K. Fujio, E.E. Tucker, J.F. Scamehorn, J. Colloid Interface Sci. 214 (1999) 224. [26] A.J. Maher, R.E. Junger, US Patent 5134379, 1992. N.B. Clark, L.J. Maher / Reactive & Functional Polymers xxx (2009) xxx–xxx 7 ARTICLE IN PRESS Please cite this article in press as: N.B. Clark, L.J. Maher, , React. Funct. Polym. (2009), doi:10.1016/j.reactfunctpolym.2009.03.011 . conductance of inkjet printed or screen printed polyaniline films with a radio frequency detector. The system has the advantage of non-contact detection of ammonia. Non-contact, radio frequency detection of ammonia with a printed polyaniline sensor N.B. Clark a, * ,