Article pubs.acs.org/ac
DC Magnetron Sputtered Polyaniline-HCl Thin Films for Chemical Sensing Applications Nicola Menegazzo,† Devon Boyne,† Holt Bui, Thomas P. Beebe, Jr., and Karl S. Booksh* University of Delaware, Department of Chemistry and Biochemistry, Newark, Delaware 19716, United States S Supporting Information *
ABSTRACT: Thin films of conducting polymers exhibit unique chemical and physical properties that render them integral parts in microelectronics, energy storage devices, and chemical sensors. Overall, polyaniline (PAni) doped in acidic media has shown metal-like electronic conductivity, though exact physical and chemical properties are dependent on the polymer structure and dopant type. Difficulties arising from poor processability render production of doped PAni thin films particularly challenging. In this contribution, DC magnetron sputtering, a physical vapor deposition technique, is applied to the preparation of conductive thin films of PAni doped with hydrochloric acid (PAni-HCl) in an effort to circumvent issues associated with conventional thin film preparation methods. Samples manufactured by the sputtering method are analyzed along with samples prepared by conventional dropcasting. Physical characterization (atomic force microscopy, AFM) confirm the presence of PAni-HCl and show that films exhibit a reduced roughness and potentially pinhole-free coverage of the substrate. Spectroscopic evidence (UV−vis, FT-IR, and X-ray photoelectron spectroscopy (XPS)) suggests that structural changes and loss of conductivity, not uncommon during PAni processing, does occur during the preparation process. Finally, the applicability of sputtered films to gas-phase sensing of NH3 was investigated with surface plasmon resonance (SPR) spectroscopy and compared to previous contributions. In summary, sputtered PAni-HCl films exhibit quantifiable, reversible behavior upon exposure to NH3 with a calculated LOD (by method) approaching 0.4 ppm NH3 in dry air.
T
gated.10−12,14,15 However, poor solubility of the polymer in most solvents gives rise to limited control over the thickness and topography of thin films formed by these techniques. Solvent incorporation within the polymer network may also be problematic and affect the polymer’s characteristics.2,10,14,16 Though polymer dissolution is not an issue with electrochemical techniques, as it relies upon localized formation of the polymer from the solvated monomer, the technique is only accessible to electrically conducting substrates.17 Less conventional approaches, such as pulsed laser deposition,10,18 pulsedplasma polymerization,4,19 ionized and neutral cluster beam deposition,10,20 thermal evaporation,21−24 and ion sputtering25 rely on vacuum deposition and aim at superior thickness control and topographical uniformity. Despite requiring elaborate and oftentimes expensive instrumentation, exquisite control over the film’s physical and chemical properties becomes readily accessible. Vacuum deposition of PAni, therefore, displays the potential to simplify integration of chemically responsive films into (micro)fabrication techniques for (e.g.) electrochromic and microelectronic devices,26,27 and, of particular interest to our research focus, miniaturized chemical sensors.28,29 DC sputtering is a vacuum deposition
he versatility in chemical functionality, physical properties, and processability of polymeric thin films has led to their implementation into a variety of biomedical, protective (corrosion or abrasion inhibition), and thermally/electrically insulating applications.1−3 In particular, electrically conducting polymers have been employed in optoelectronic devices, photovoltaic cells, microelectronics, and chemical sensors, among others.2,4 Distinct changes in optical and electronic properties and temporal stability as well as quasi-metallic conductivity render the emeraldine salt form of polyaniline (PAni) the most practical for integration into various functional devices.5−8 Depending on the dopant and the arrangement of repeating benzenoid and quinoid units, PAni can exists in different forms with the relative fraction of each of these units dictating the oxidation state. The fully reduced form, leucoemeraldine, consists solely of benzenoid amine units whereas the fully oxidized form, pernigraniline, is exclusively composed of quinoid imine units. A neutral oxidation state contains equal amounts of both units and is referred to as the emeraldine base.9−13 Subsequent protonation, or doping, of the base yields the electrically conductive emeraldine salt counterpart.9,11,13 The chemical and physical properties of thin films based on the emeraldine salt vary depending upon the preparation process (synthesis and deposition).9 Several methods, including drop- and spin-cast deposition, as well as chemical and electrochemical polymerization have been investi© 2012 American Chemical Society
Received: April 19, 2012 Accepted: June 6, 2012 Published: June 6, 2012 5770
dx.doi.org/10.1021/ac301006f | Anal. Chem. 2012, 84, 5770−5777
Analytical Chemistry
Article
technique applied to formation of thin films from electrically conducting target materials and as such, compatible with doped PAni. In addition, since the resulting thin films are deposited directly in the conducting state, the need for postdeposition doping is bypassed, streamlining integration into procedures relying on cleanroom environments. This contribution focuses on the implementation of DC magnetron sputtered PAni doped with hydrochloric acid (PAni-HCl) and subsequent application of as-deposited thin films to chemical sensing. The ultimate scope being facilitated integration of chemically interacting polymers into microfabricated sensors. Detection of gas-phase NH3 was selected as a test application for this novel material due to interest in measuring ppm levels in a variety of fields, including environmental monitoring, human health, and industrial hygiene.30−32 Furthermore, our group previously reported efforts demonstrating the feasibility of surface plasmon resonance (SPR) spectroscopic sensing of NH3 at low-ppm levels utilizing conventional drop-cast doped PAni.33 Therefore, the viability of sputtered PAni-HCl thin films for chemical sensing can be directly verified against baseline performance measurements previously established in-house as well as external reports available in the literature. Topographical characterization by atomic force microscopy of sputtered films show highly uniform coverage of the substrate with significantly lower surface roughness compared to drop-cast counterparts. Spectroscopic analysis with Fourier transform infrared (FT-IR), UV−vis, and X-ray photoelectron spectroscopy (XPS) indicate that chemical rearrangement of the carbon network, decreased protonation, and increased proportion of covalently bound Cl occurs during the sputtering process. In spite of these side-reactions, deposited films retain sufficient electrical conductivity to enable gas-phase detection of NH3 in the low-ppm range.
solution is strongly oxidizing and should be handled with great care!] for at least 60 min, followed by copious rinsing with deionized water and ultrasonication in a 5:1:1 solution of ultrapure water, NH4OH, and 30% H2O2 (Fisher Scientific, Fair Lawn, NJ) for an additional 60 min. The slides were then rinsed, sonicated several times with ultrapure water, and finally dried with a stream of dry nitrogen gas (Keen Compressed Gas Co., Wilmington, DE). Quartz substrates (Electron Microscopy Sciences, Washington, PA) used for UV−vis studies were cleaned in a similar fashion. PAni-HCl layers were deposited with a Cressington DC magnetron sputtering system (model 308R, Cressington Scientific Instruments Ltd., Watford, UK). For compositional and gas-phase analysis, polymer films were prepared on goldcoated glass slides (5 nm Cr, 99.95+%, Kurt J. Lesker, Clairton, PA followed by 50 nm Au, 99.99%, Espi Metals, Ashland, OR). A copper disk was secured between the cathode and the target to increase the electrical contact between the two, improving sputter efficiency and reducing resistive heating of the target. Reaching the required deposition pressures with new targets required additional time due to outgassing of the polymer. Cycling the pressure in the deposition chamber between 10−2 and 1 Pa, achieved by controlling the amount of the sputtering gas (Ar, Keen Compressed Gas Co., Wilmington, DE) leaked into the system, helped to improve the quality of the deposited layers by diluting interfering atmospheric species. Each sample was positioned 35 mm from the target and sputtered with a current density of 12 μA mm−2 at a pressure of 10−3 to 10−2 Pa. Visual inspection of the resulting films sputtered on glass displayed a greenish tint, the color expected for PAni-HCl.11 PAni-HCl was also drop-cast from a 1 mg mL−1 solution of PAni-HCl dissolved in formic acid (Acros Organics, NJ, USA) from the same emeraldine salt powder used for the sputtering targets. In order to promote solvation, the mixture was sonicated for ∼2 h prior to drop-casting onto quartz, gold, and glass substrates. Characterization. Characterization was performed in tandem for drop-cast and sputtered samples. Absorbance measurements of the coated quartz substrates were collected using a Hewlett-Packard spectrophotomer (model 8452A, Santa Clara, CA) in transmission mode at a 2 nm resolution. For the gold coated substrates, mid-IR spectra were acquired using an AutoSeagull specular reflectance accessory (Harrick Scientific, Pleasantville, NY) operated at an incident angle of 87° in a Bruker Optics Vertex 70 (Billerica, MA) FT-IR spectrometer equipped with a mercury−cadmium−telluride detector. Each spectrum consisted of 100 averaged scans collected over the 400−4000 cm−1 range. XPS surface composition was determined using an ESCAlab 220i-XL electro spectrometer (VG Scientific, UK) with a monochromatic aluminum Kα (1486.7 eV) X-ray source. The nominal spot size for measurements was 400 μm (80−20) operated at 15 kV with a power of 100 W. Survey spectra were collected with an energy resolution of 1 eV from 0 to 1200 eV binding energy with a pass energy of 100 eV and a dwell time of 100 ms per data point. High-resolution spectra were collected with energy resolution of 0.1 eV with a pass energy of 20 eV and a dwell time of 100 ms per data point. Signal averaging was used to improve the signal-to-noise for the high-resolution spectra. Spectral deconvolution was performed with CasaXPS Peak (v. 2.3.26) by separating each high resolution peak into four Gaussian-shaped contributions centered in accordance with literature values. Polymer thickness and surface roughness were
■
MATERIALS AND METHODS Target Preparation. In contrast to previous work utilizing camphorsulfonic acid (CSA) as the dopant, it was anticipated that CSA might decompose during the sputtering process introducing additional complexity into the system being studied; therefore, hydrochloric acid was preferred for this initial contribution. The conductive emeraldine salt was obtained by initially ensuring full deprotonation of the emeraldine base (MW ∼5000, Sigma-Aldrich, St. Louis, MO). This was achieved by stirring the emeraldine base in 1 M NH4OH (Fisher Scientific, Fair Lawn, NJ) for 1−2 days followed by gravity filtration. The filtrate was then suspended in 2 M HCl (Fisher Scientific, Fair Lawn, NJ), for 3 days, and the emeraldine salt was filtered and dried for several days. Aliquots of the dried PAni-HCl powder weighing 1.25−1.50 g were compressed using a custom-made die with up to 16 kg/mm2 of pressure from a hydraulic press, forming circular sputtering targets of 38 mm diameter and ∼1.5 mm thick. A thin copper mesh was embedded into the powder prior to compression, in order to provide the target with additional mechanical strength. Without the support mesh, the PAni-HCl targets crumbled in the plasma during the initial attempts to sputter. Co-sputtering of Cu was prevented by limiting the amount of samples sputtercoated with each target. Sample Preparation. Circular glass coverslips (Fisher Scientific, Pittsburgh, PA) with a diameter of 25 mm were cleaned with heated piranha solution [3:1 (v/v) 70% H2SO4/ 30% H2O2 (Fisher Scientific, Fair Lawn, NJ) - Caution: piranha 5771
dx.doi.org/10.1021/ac301006f | Anal. Chem. 2012, 84, 5770−5777
Analytical Chemistry
Article
Figure 1. (A) UV−vis and (B) mid-IR absorption spectra for drop-cast and sputtered PAni-HCl.
at best. Leucoemeraldine, the fully reduced form of polyaniline devoid of quinoid groups, displays very similar spectral features36 indicating that DC sputtered PAni-HCl may undergo structural changes during the deposition process. Comparable sloping spectra have been presented by Lim et al.10 for thin films deposited by laser ablation of a PAni target at high fluence wherein significant rearrangement of the carbon network and loss of hyperconjugation were reported. Similar spectra have also been observed with plasma polymerized PAni.37 Figure 1B depicts FT-IR spectra for drop-cast and sputtered PAni-HCl, respectively. Detailed peak assignment is summarized in Table 1, Supporting Information. While the FT-IR spectrum of the drop-cast samples resembles those available in the literature,38−40 several noticeable differences appear in sputtered PAni-HCl suggesting that, although the basic carbon functionality is partially retained, disruption of hyperconjugation is possible during the sputtering process. The presence of the “free carrier tail”, characterized as the increasing baseline absorbance in the spectrum of drop-cast samples, is assigned to delocalization of electrons in highly conducting samples.41−43 The absence of this tail in sputtered PAni-HCl suggests a lower fraction of charge carriers and a loss in electrical conductivity. Similarly, the peak centered at 1176 cm−1, generally associated with the presence of positive charges and electron delocalization as well as NquinoidN stretches,39,40 displays a decreased intensity in sputtered samples compared to dropcast counterparts. Comparison of the relative peak areas for the spectral features corresponding to quinoid and benzenoid units showed that sputtered samples are richer in benzenoid amines, whereas drop-cast samples display relative peak areas closer to the 50:50 arrangement expected for the emeraldine form. In accordance with UV−vis analysis, a higher fraction of benzenoid suggests that the structure of sputtered PAni-HCl may resemble that of leucoemeraldine.9 Additional evidence is provided by the peaks at 1313 and 1246 cm−1 in the spectra of sputtered samples which are consistent with increased fraction of benzenoid− benzenoid−benzenoid (BBB) ring arrangement. Furthermore, the broadness of the peak centered at 1329 cm−1 in the spectrum for drop-casted samples arises from combination of the different vibrational modes of CN for different ring arrangements.44 Contributions from different ring arrangements in the sputtered spectrum are much lower, overall yielding a sharper peak with a shoulder.
investigated by atomic force microscopy (AFM, Bruker AXS, model Dimension 3100, Santa Barbara, CA) in tapping mode with silicon probes (BudgetSensors, model Tap300, Sofia, Bulgaria). Prior to measuring surface roughness, micrographs were flattened with a first-order polynomial fit routine provided by the instrument manufacturer. Ammonia Sensing. Gas-phase ammonia sensing with the sputtered samples was performed using the SPR assembly previously described by Menegazzo et al.33 following minor adjustments. Specifically, the gas delivery was reconfigured in order to mitigate adsorption of the analyte onto the tubing walls, resulting in placement of the MFC regulating NH3 delivery next to the flow cell, rather than at the stock gas tank. This rearrangement resulted in a decrease of exposed tubing from 2.5 m to less than 0.2 m. Following an equilibration period with dry air of at least 60 min, the analytical cell was exposed to 10 min of diluted ammonia followed by 30 min of air; this was cycled in triplicate for four concentrations (80, 192, 300, and 512 ppm). Each spectrum consisted of 5 averaged scans to improve signal-to-noise ratios. For λspr = 546, scan averaging was increased to 8 to compensate for a broader dip resulting from overlapping absorption features from the polymer (Figure 1A).33 In order to account for fluctuations independent of the NH3 (e.g., temperature shift, spectrometer drift), the reference signal was subtracted from the analytical signal.
■
RESULTS AND DISCUSSION Spectroscopic Characterization. The absorption spectrum was recorded for both sputtered and cast PAni-HCl films in the visible region (Figure 1A). Consistent with literature,34,35 the drop-cast samples display distinct absorption peaks characteristic of the emeraldine salt films cast from formic acid, including a doublet at 316 and 434 nm corresponding to π−π* excitation of the conjugated ring system and the polaron−π* transition, respectively, as well as a broad band peaking at approximately 800 nm (only partially resolved due to instrumental cutoff) which is normally associated with π− polaron transition. Similarly identifiable peaks were not observed for sputtered PAni-HCl spectra; instead, a single monotonically decreasing absorption feature covering the spectral range examined was observed. Upon closer inspection, a shoulder at ∼280 nm may be present which has been attributed to the π−π* transition of the ring system;4 however, the weak absorption of this band renders assignment tentative 5772
dx.doi.org/10.1021/ac301006f | Anal. Chem. 2012, 84, 5770−5777
Analytical Chemistry
Article
Figure 2. Atomic force micrographs of (A) sputtered PAni-HCl and (B) drop cast PAni-HCl over a 5 × 5 μm2 area.
X-ray Photoelectron Spectroscopy. Examination of the survey spectra (N 1s and Cl 2p) for both sample types (Figure 1A, Supporting Information) confirm the absence of copper, indicating that cosputtering of the embedded mesh did not occur. Decomposition of the high resolution N 1s and Cl 2p peaks was performed by utilizing four components (for each peak) at binding energies consistent with literature,9,37,45 and the results have been summarized in Table 1 with exemplary spectra shown in Figure 1B−E as Supporting Information.
The increase in benzenoid amine units may be attributed to structural changes brought on by fragmentation and radical formation during the sputtering process. Breaking of polymer chains can initiate cross-linking events induced by electron collisions.4 This is further shown by the presence of the peaks at 752 and 698 cm−1 in the sputtered spectrum, commonly attributed to ring deformation,43 and typically appear in PAniHCl samples subjected to repeated protonation/deprotonation cycles and in the presence of oligomers in the polymer mixture.39,43 These characteristics are consistent with a high degree of cross-linking and/or an increased presence of terminal monosubstituted phenyl ring, the latter indicating shorter polymer chains.43 The absence of these features in the drop-cast sample is not surprising since the powder used to fabricate the sputtering targets and drop-cast solutions was subjected to a single deprotonation cycle. Finally, band broadening in the 3500 to 2000 cm−1 region as well as the appearance of a nitrile peak at 2210 cm−1 further confirm that structural changes consistent with a more reduced and crosslinked/branched form of PAni-HCl were introduced in the polymer network upon sputtering. Similar changes have been observed for thermally evaporated films.21,45,46 Surface Analysis. Atomic Force Microscopy. The sputtering rate of PAni-HCl and topography of the resulting layers were characterized by AFM (Figure 2A); micrographs of drop-cast samples were also measured for comparative purposes (Figure 2B). The thickness of the polymer layers was determined by masking a portion of the substrate and measuring the resulting step size (25 ± 3 nm). The deposition rate of PAni-HCl was calculated to be approximately 0.04 nm sec−1, which is well within the range reported for other vacuum deposition techniques such as thermal evaporation,23,47 cluster beam and laser ablation,10 and plasma polymerization.4,48,49 Sputtered films show highly uniform, nearly featureless, coatings similar to those obtained by plasma polymerization,49 a technique commonly used to produce pinhole-free films,50,51 though small agglomerations are occasionally observed to protrude from the surface (see top of Figure 2A). Although their exact origin is presently unknown, it is possible that they may correspond to either clusters of the polymer fragmenting from the target or atmospheric contaminants (e.g., dust particles) collected at the surface postdeposition. Much like films obtained by pulsed laser deposition,10 the surface roughness (Rrms) of sputtered samples is substantially lower compared to drop-cast samples of comparable thickness: 0.5 nm compared to 20.4 nm, respectively.
Table 1. Summary of the Binding Energy and Relative Abundance for the N 1s and Cl 2p Phoemission Peaks for Sputtered and Drop-Cast Films drop-cast N/Cl (%) binding energy (eV) Cl 2p ionic3/2 ionic1/2 covalent3/2 covalent1/2 N 1s quinoid imine benzenoid amine protonated amine protonated imine
197.3 198.8 200.1 201.8
± ± ± ±
0.1 0.2 0.1 0.1
sputtered
58:42
71:29
rel. abundance (%)
rel. abundance (%)
38 37 21 4
± ± ± ±
1 1 3 2
7 7 69 18
± ± ± ±
1 1 2 1
398.3 ± 0.2 399.5 ± 0.2 400.8 ± 0.1
0.2 ± 0.2 60 ± 3 31 ± 3
14 ± 4 64 ± 5 21 ± 1
402.3 ± 0.3
8±2
1±1
Drop-cast layers displayed near-baseline levels of neutral imine sites in agreement with published reports on emeraldine salt layers;9,52 decomposition of the N 1s peak into the individual components suggests that the overall structure of the polymer is more accurately represented as the semiquinone radical salt, which is composed of intermittent amine radical cations and neutral amines with residual cationic quinoid units.38,53 The sputtered films appear to have retained a similar amount of neutral benzenoid units as the drop-cast films, while the neutral quinoid component increased at the expense of the protonated units. Comparison of the peak areas for the neutral and charged fractions in drop-cast films show an approximate 0.67 ratio ([N+]/[No]), while the sputtered films show a decreased protonated-to-neutral ratio of ∼0.28 which may result, at least in part, from neutralization by free electrons present in the plasma. The increase in neutral imine and amine groups is in agreement with the notion that rearrangement of 5773
dx.doi.org/10.1021/ac301006f | Anal. Chem. 2012, 84, 5770−5777
Analytical Chemistry
Article
Figure 3. (A) SPR sensorgram for sputtered PAni-HCl for λSPR at 620 nm and (B) linearized calibration for the sensor at the λSPR tested for different NH3 concentrations (80, 192, 300, and 512 ppm) corrected for background drift. X-axis (concentration) error bars are manufacturer tolerance for the mass flow controller; Y-axis error bars are based on replicate measurements and inherently include the propagated effects of mass flow controller variability.
between incident photons and surface plasmon polaritons.59 The shallow penetration depth (
