Texturing of nanocoatings for surface acoustic wave-based sensors for volatile organic compounds

1 Introduction

The predictive control of the surface morphology of gas sensing films is very important for the sensor performance, especially in the cases, when the analyte detection mechanism is surface adsorption. A morphology that provides a strongly revealed surface, or increased specific area, has been found favorable for enhancement of the gas-sensitive behavior and improvement of the sensor parameters [1–3]. The application of a method for surface area increase, which is compatible with the conventional microfabrication technology flow has been crucial for the development of a low-cost, but at the same time precise sensor solution. An acceptable ratio of price to quality could be achieved if nanopatterning or nanolithographic techniques are avoided. Recently, a lot of examples of precise, but expensive, or functional area limited approaches for coatings texturing have been available in the literature [4–7]. Most often, the patterned, or textured coatings are of metal-oxides and the gas detection principle is based on the chemical absorption of the target molecules, resulting in a resonance frequency change (quartz crystal microbalance, or cantilever sensor), conductivity (resistance) change, or threshold voltage modulation.

Over the past decade, carbon-based macromolecules, and especially conductive polymers, have attracted the attention of researchers in the field of gas sensor technology due to the low-cost film deposition methods, such as spin-coating, dip coating, spray deposition, ink printing, etc. [8]. The physical and chemical properties of carbon-based materials strongly depend on their spatial structure and controlled morphology. There have been numerous recent reports proving that the patterning of coatings based on carbon nanomaterials significantly affects the coating’s functionality, and can essentially improve their basic characteristics despite the concrete electronic device where they have been implemented [9–11].

Carbon atoms can be organized in different types of hybridization such as graphite, graphene, carbyne, polymers, and other allotrope forms. Especially attractive has been the conjugated polymer poly(3,4-ethylene dioxythiophene) doped with polystyrene sulfonate (PEDOT:PSS), because it is conductive and the adsorption of specific organic molecules on its surface has resulted in an electrical resistance change, making it suitable for the design of chemoresistors [12]. Its chemical structure, electroconductive properties, can be easily tailored, making it a flexible material applicable in a great variety of electronic devices, including detection of volatile organic compounds. However, the resolution of the resistivity change is not as precise as the nowadays requirements (corresponding to the response to the smallest change of the concentration beyond ppm). Thus, there is an urgent need for a more sensitive approach for the precise detection of volatile organic compounds by using PEDOT:PSS.

Recently, advanced approaches to enhance the properties of the nanomaterials and improve the selectivity of the sensors have been reported. Among them, can be highlighted alloys decoration onto 3D graphene network [13], three-dimensional porous reduced graphene oxide decorated with carbon quantum dots and platinum nanoparticles [14], and nitrogen/sulfur co-doped graphene [15]. However, these approaches can be ascribed to the chemical processes and the sensors utilizing these nanomaterials are of electrochemical type of operation. Surface acoustic wave (SAW) can be a suitable platform for the realization of precise microsensors, due to the principle of operation relying on the change of the parameters of a SAW generated and distributed on the sensor’s substrate interdigital transducers (IDTs) have been patterned at the input and output of the sensor. Over them, a piezoelectric layer has been deposited, or the substrate has been a piezoelectric crystal that is responsible for the travelling of the acoustic wave from the input to the output with a certain delay. When a gas-sensitive coating has grown on the path of the wave, each adsorbed molecule has resulted in a mass change of the coating, therefore in a change of the resonance frequency of the SAW sensor with a resolution, corresponding to a nanogram change in the concentration of the analyte [16]. This method is superior over the chemical approach, as it relies mainly on physical interaction between the waves and the sensing nanoparticles and in most of the cases between the sensitive nanoparticles and the analyte of interest (i.e., van der Waals adsorption). In addition to the excellent sensitivity, SAW sensors do not require threshold voltage to overcome as in the semiconductor sensors, and their fabrication is simple with two or three photomasks.

Although, to the best of the author’s knowledge, there is no direct evidence reported in the literature for texturing ability of PEDOT:PSS by using this approach, such assumption was made during the design of the current experiment. It was based on the suggestion that texturing is possible if the SAW parameters comply with the molecules’ sizes. There have been sufficient data on the selection abilities of some IDTs, moving specific particles and extracting them among others like a filter. Except for a separation, the mixing of particles has been also possible due to their displacement, caused by the acoustic waves with a specific velocity that can be governed by the geometry of the IDTs [17,18].

The goal of this study was to achieve texturing of a gas-sensitive film avoiding expensive techniques and lithographic patterning. In this way, it was expected enhancement of the performance of precise sensors due to the increased specific area of the nanocoating induced during its growth. It was also aimed for the development of a low-cost, large-area method for texturing and expanding the adsorption capacity, which could be applicable in a variety of sensing structures, no matter if they are optical, mechanical, biological, etc. The novelty of the work is in the approach of texturing by using the propagation of SAW with specific parameters during the growth of the sensing film, resulting in its patterning, which is reported for the first time for the PEDOT:PSS polymer.

2 Methods

Following a well-known methodology for SAW design, a suitable device was produced with specific geometry of the interdigital electrodes to keep the wave unquenched during the sensing film deposition [19]. The finger length of the interdigital electrodes was 150 µm, the finger pitch was 25 µm, and the number of fingers in each transducer from the pair (input and output IDT) was asymmetric and equal to 4 and 100, respectively (Figure 1). Glass substrates were cleaned with a solution consisting of ammonia, hydrogen peroxide, and water in a weight ratio of 1:1:3. The cleaning procedure was conducted in a heated cleaning solution for 15 min, followed by drying over vapors of isopropyl alcohol. Silver films for IDTs were deposited by vacuum DC sputtering at a base pressure of 10^–5 Torr. Conventional photolithography and wet chemical etching in potassium iodide solution were applied for the patterning of the comb-shaped electrodes.

Figure 1

Microscopic image of the part of the SAW sensor with PEDOT:PSS sensing nanocoating.

The input signal parameters for the SAW excitation were set from the AC generator as follows: frequency of 100 kHz and AC voltage of 20 V. The piezoelectric film, which is necessary to generate the acoustic wave was poly(vinylidene fluoride-trifluoroethylene) (P(VDF-TrFE)) with a thickness of 450 nm and the gas-sensitive coating was PEDOT:PSS with a thickness of 50 nm. P(VDF-TrFE) powder was dissolved in methyl ethyl ketone (MEK) solvent and the PEDOT:PSS was commercially available as a water solution (4.0% in H₂O, high-conductivity grade from Sigma Aldrich). The piezoelectric coefficient d₃₃ of the P(VDF-TrFE) was 23 pC/N and the dielectric constant was approximately 14 at room temperature and up to the operational frequency of 100 kHz. The PEDOT:PSS was selected as it exhibited good sensitivity to ethanol and can be dissolved in non-organic solution, keeping from dissolving the sub-layer during the deposition. The precursor solutions were subsequently spray-coated on the IDTs patterned substrates by fine nozzle having diameter of 2 µm, at a pulverization pressure of 4 bar at relatively low temperature of 80°C for the two precursor solutions. The piezoelectric material was spray coated all over the surface of the sample, except over the square contact pads, while the gas-sensitive coating was pulverized on the intermediate region between the input and the output IDTs. The specified areas were located by stencil masks. The depositions were conducted on a control sample without excitation and on a SAW excited sample at the moment of the PEDOT:PSS nanocoating deposition. The testing of the two samples was conducted in a conventional chamber for gas sensors electrical characterization with control of the temperature and the concentration of the ethanol vapors in the volume. The chamber was equipped with a reference ethanol sensor for calibration. Details for the testing setup can be found elsewhere [20]. Images from scanning electron microscope (SEM) and atomic force microscope (AFM) are shown to demonstrate the effect of the acoustic wave on the film’s morphology. 3D topographies of the film surfaces were scanned by AFM MFP-3D, Asylum Research, Oxford Instruments in non-contact mode, and scanning electron microscopy was conducted by microscope Philips 515. The data about the root-mean-square (RMS) roughness of the films were extracted by WSxM software. The conducted research was not related to either human or animal use.

3 Results and discussion

Figures 2b and 3b show the visible surface texturing and increased surface roughness, after the PEDOT:PSS deposition on the SAW excited substrate. For comparison, the surface of the film without texturing is also shown (Figures 2a and 3a). As can be seen, the sensing coating was uniform and consisted of wire-like particles of the conjugated polymer, which are however randomly distributed and cannot be distinguished as separate particles. The SAW distribution caused collection of the polymeric macromolecules in groups, according to their similarity in the length or diameter (approximate length of 1 µm and diameter 0.25 of the length), making them distinguishable. The RMS roughness of the non-textured film was 6.6 nm vs 12.8 nm for the textured film. This molecules assemble change, induced with the excitation, was ascribed to the SAW interaction with the PEDOT:PSS molecules and the energy transfer, during this interaction, which seemed to control the position of the particles until reduction of the free surface energy. It is expected to facilitate the adsorption process for the analyte molecules.

Figure 2

SEM images of PEDOT:PSS coating: (a) non-textured and (b) textured during its growth on the SAW excited substrate.

Figure 3

AFM 3D profile of PEDOT:PSS coating: (a) non-textured and (b) textured during its growth on the SAW excited substrate.

The morphology affected the main characteristics of the SAW sensor, like sensitivity, dynamic range, reproducibility (hysteresis), response, and recovery time. Figure 4 represents the measured results for the textured film, and Figure 5 – for the non-textured. The sensor response was measured as an output voltage attenuation with the vapor concentration increase. It is a synchronized process with the time delay of the output voltage as compared to the reference measurement channel. A stable trend in the output signal variation with the change of the adsorbed analyte was observed in Figure 4a.

Figure 4

SAW sensor response to ethanol: (a) output voltage and time delay between the output and input voltage of IDTs, (b) hysteresis, and (c) response and recovery times for the textured PEDOT:PSS coating.

Figure 5

SAW sensor response to ethanol: (a) output voltage and time delay between the output and input voltage of IDTs, (b) hysteresis, and (c) response and recovery times for the non-textured PEDOT:PSS coating.

The sensitivity was 5 µV per ppm with a dynamic range from 0 to approximately 7,000 ppm, which is a broad dynamic range. It was ascribed to the increased surface area after texturing, which facilitates the diffusion of the analyte molecules deeper in the sensitive nanocoating, escaping the surface for new incoming ethanol molecules to be adsorbed. The hysteresis (reversibility of the sensor response at a gradual decrease of the concentration), shown in Figure 4b, was 10 mV that introduces a measuring error of 0.6%, which is acceptable for such measurements and together with the dynamic range and the sensitivity, it is in line with the current standards for ethanol sensors using other operation principles [21,22]. The response time was 970 ms and the recovery time was longer (3 s) due to the mechanism of desorption without additional thermal energy, because the sensor operates without a heater (Figure 4c). The relatively fast response time can be explained by the fast diffusion of the analyte molecules. Since the overall thickness of the film was not increased, but only the surface was textured, the overall surface-to-volume ratio increases, making fluent motion of the detected particles. In contrast to these results, the sensing structure with twice flatter non-textured surface gave a linear and stable response after 5,000 ppm, which corresponded to a narrow dynamic range (only 2,000 ppm vs 7,000 ppm for the textured sensing structure). Although the sensitivity in the linear region was 12 µV per ppm, which is more than twice higher than for the rougher coating, the adsorption capacity was too small (Figure 5a). The hysteresis was greater (Figure 5b) and introduced a measuring error of 4%, which is not acceptable for the ethanol vapor detection. The response time of 1.1 s was relatively close in value to the textured film because of the same adsorption van der Waals mechanism, but the recovery time was faster (1.31 s) due to the easier desorption of the analyte molecules from the flat surface of the sensing film (Figure 5c).

4 Conclusions

To control the texture of the gas/vapor sensitive nanocoating, an asymmetric topology of the IDT electrodes was used for generating SAWs with specific space distribution; therefore, inducing electrical fields with specific parameters, which in turn caused the assembling of the nanoparticles building the sensitive coating. The SAW sensor with the textured nanocoating exhibited revealed surface and enhanced dynamic range to ethanol with good sensitivity, small hysteresis, fast response time, and acceptable recovery time. The advantages of the proposed method are easy gas-sensitive properties tailoring, large-area processing, implementation of conventional deposition, and compatible patterning technologies (no need for lithographic techniques for the texturing of the nanocoating). Possible applications of the fabricated sensor could be the food industry, pharmacology, and microelectronic industry, where ethanol presence is not a critical factor for the safety, but it is required for precise measurement of its vapor concentration. The major limitation of the study is the impossibility of unidirectional alignment of the patterned molecules, which is related to the electrodes’ topology, as well as the susceptibility of patterning of the PEDOT:PSS material, which is related to the length of the molecule. Possible routes to improve these limitations are by using additional pair of comb electrodes located perpendicular to the main IDTs, which will however increase the complexity of the fabrication, and replacement of the sensitive material with another molecule, which might deteriorate the response time of the sensor and its susceptibility to interferences. Future work will be related to studying the selectivity of the sensor toward a mixture of similar substances that will probably need a more complex design of the IDT electrodes.