Environmental Monitoring of Hydrocarbons: A ... - ACS Publications

Robert Stach , Bobby Pejcic , Emma Crooke , Matthew Myers , and Boris Mizaikoff ..... Bobby Pejcic , Matthew Myers , Nilukshi Ranwala , Leigh Boyd , M...
0 downloads 0 Views 139KB Size
Critical Review

Environmental Monitoring of Hydrocarbons: A Chemical Sensor Perspective BOBBY PEJCIC,* PETER EADINGTON, AND ANDREW ROSS CSIRO Petroleum, ARRC, P.O. Box 1130, Bentley, WA, 6102, Australia

Assessing the environmental impact of organic pollutants requires reliable analytical tools that can rapidly screen them with minimal sample handling. Chemical sensors are expected to play an increasing role in environmental monitoring, and recent technological advances are certain to facilitate the application of chemical sensing devices. The search for highly selective, sensitive, low cost, stable, and robust sensors for hydrocarbons is an area of interest that is reflected by many publications on this topic. This report surveys some of the work that has been undertaken using sensors to detect hydrocarbons in the gas and liquid phase. The analytical capabilities of various sensors are compared and discussed in terms of their selectivity, sensitivity, and detection limit. It was found that the sensitivity is highly dependent on the experimental conditions used in the preparation of the sensing surface. Many sensors display acceptable sensitivity under controlled laboratory conditions; however, very few are selective enough to distinguish among several hydrocarbons in complex mixtures. Selectivity is still a challenge that is hindering the widespread application of chemical sensors for environmental monitoring of hydrocarbons and a number of strategies have been proposed to help overcome some of these problems.

Introduction The emission of organic pollutants into the environment, either through man-made or natural processes (1-3), is a topic that generates considerable scientific interest and public concern. The projected increase in global energy use over the next several decades is expected to facilitate unwanted pollutants and this will be a significant challenge for governments attempting to reduce levels without stifling economic growth (4). Although a number of methods have been proposed to tackle this controversial issue (5), to really understand and mitigate the impact of environmental contaminants requires sophisticated tools that can provide information about their activity, toxicity, and cycling. The desire to monitor all adverse chemicals in our environment has been the impetus behind the growth of many innovative tools over the last century. Various methods are available for screening and assessing the impact of hydrocarbon-based pollutants on the biosphere. Traditionally, qualitative and quantitative analyses are performed using standard analytical techniques such as gas/liquid chromatography and mass spectrometry (6). A sample is collected and brought back to the laboratory where * Corresponding author e-mail: [email protected]. 10.1021/es0704535 CCC: $37.00 Published on Web 08/18/2007

Published 2007 by the Am. Chem. Soc.

it undergoes various lengthy separation/preconcentration steps. Despite being highly sensitive and reliable, these techniques still suffer from miniaturization problems and are not readily amenable for the real-time monitoring of pollutants in the field. Consequently, there is a great deal of interest in developing alternative analytical technologies, noting that chemical sensors are an attractive tool which allow rapid on-site and in-situ monitoring of environmental contaminants. The two most important measurable qualities of a chemical sensor are its selectivity and sensitivity, and a great deal of effort has been directed toward characterizing the sensor in terms of these parameters. For analytical purposes, a sensor that responds directly to a single molecule of interest is highly desirable. However, most environmental samples, regardless of their source, comprise a range of hydrocarbons making the detection of individual components a challenge, which pushes any analytical device to its limits. Furthermore, the presence of ionic salts and humic compounds in natural waters may interfere with the sensor response, so that their removal or separation is usually necessary. In fact, reports suggest that humic substances have a strong propensity to attach onto the sensor surface (7). By contrast, sensors for volatile organic compounds in air usually encounter fewer problems in terms of fouling; however, other issues such as humidity and temperature play an important role on sensor sensitivity (8, 9). Evidently, hydrocarbon pollution is widespread and the concentrations observed in the atmosphere can vary between sub-µg/L and sub-mg/L (10); whereas the hydrocarbon levels found in natural waters are normally lower (ng/L to µg/L) (11). There are many factors that regulate the hydrocarbon activity in the environment and these can vary significantly from one region to another (12). Obviously a sensor that responds linearly to hydrocarbons over a wide concentration range (ng/L to mg/L) with a detection limit of ∼0.1 ng/L would be ideal. Likewise, the sensor response time and reversibility are other important considerations. A sensor that takes several minutes to respond may be inadequate for real-time monitoring. In addition, sensor drift is an issue that also requires attention if long-term continuous measurements are to be performed. Clearly the sensor must be robust and able to withstand some of the harsh conditions that may be encountered in the natural environment (salinity, humidity, temperature variations of 0-50 °C, etc.). The relative importance of the above factors will in many ways depend on the type of environment and monitoring necessities. There is an abundance of data on the distribution and concentration of hydrocarbons in the biosphere; however, our understanding is somewhat limited since the natural processes governing chemical species behavior and dynamics VOL. 41, NO. 18, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

6333

FIGURE 1. Research papers published on hydrocarbon sensors over the past two decades.

FIGURE 2. Materials used as receptors in the design of hydrocarbon sensors. Note that the group “others” comprises inorganic metal oxides, semiconductors, and organic films, etc. is complex. It is not surprising that the development of sensors for the detection of hydrocarbons has steadily increased since 1990 (see Figure 1). Recent reviews (13) of the subject have mainly focused on the detection of methane, while others (14) have discussed the use of sensors for the environmental monitoring of certain inorganic and organic gas analytes. Consequently, it seems timely to review the advances in this exciting branch of analytical and environmental science. Since many articles have been published in the area of chemical sensors, this manuscript is not meant to be a comprehensive review of the field, but rather a critical review, presenting a selection of the most significant technologies and advances in relation to environmental monitoring of hydrocarbons. It is not the intention of this review to critically evaluate each type of chemical sensor nor will it present a summary of recommendations for effective sensing. Rather, the objective of this review is to address the development of chemical sensors for the detection of aliphatic and aromatic hydrocarbons.

Chemical Sensor’ Properties The receptor and the transducer are the two most important components of a chemical sensor. The analytical features of various sensors used for hydrocarbon detection are summarized in Tables 1-3 of the Supporting Information. Normally, they are classed according to the type of transducer and the three common transduction methods used are (1) mass/piezoelectric, (2) optical, and (3) electrochemical/ electrical. It appears that the sensitivities of the various transducers are somewhat comparable and the variations observed are mainly due to the experimental conditions 6334

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 18, 2007

employed during sensor fabrication and operation. Although it is evident that the information on the sensor sensitivity and detection limit is missing in many cases. Nevertheless, most sensors do suffer from some degree of cross sensitivity, and tailoring a receptor so that it exclusively interacts with the hydrocarbon molecule of interest is a challenge for building a reliable analytical device. This review will pay particular attention to the performance of various receptors, noting that the receptor is responsible for selectivity, and this is an important issue which is hindering the deployment of sensors for environmental monitoring of hydrocarbons.

Receptors The selection of a suitable receptor material is an important step in designing a reliable sensor. A majority of the research activity has been directed toward the search for robust materials with exceptional selectivity. In principle, any material that has the ability to adsorb/absorb hydrocarbon molecules at significant levels is a potential receptor candidate. Naturally, the adsorption/absorption capacity does not only depend on the chemical structure and composition of the material but also on its physical properties. Figure 2 shows the common receptor materials used in sensors for hydrocarbon detection, noting that polymers are by far the most popular material. This is not surprising considering that they can be easily tailored to generate a material with the desired physical and chemical properties. Various groups (15-18) have developed models and showed that the response mechanism of polymers is dictated by the thermodynamics of hydrocarbon extraction/partitioning. These empirical methods work well for hydrocarbon-

polymer interactions in the vapor phase; however, very little information is available on the hydrocarbon-polymer selectivity in aqueous solutions and in the presence of hydrocarbon mixtures. Although, some reports suggest that the method of coating and polymer crystallinity play a significant role on sensor response (19, 20). It is apparent that polymers display promising analytical results; however, there are still concerns that many of them are unable to achieve the desired selectivity and sensitivity in real samples. In addition, some polymers undergo degradation or an irreversible swelling process when exposed to certain environments (21, 22), while others adsorb and respond to water vapor (23). Although, it has been revealed that the response characteristics of polymer-based chemical sensors can be improved by incorporating carbon particles (24-29), copper (30), inorganic salts (31), and sulfonate salts/ compounds (32, 33) into the polymer framework. Others have shown that molecular imprinting is an alternative method of introducing sensor selectivity (34-37). This process is a convenient way of creating host-guest materials, noting that much improved selectivities and sensitivities can be achieved when the polymerization conditions are optimized. In some cases appropriate sensor selectivity cannot be achieved either due to the complex nature of the sample or the very weak forces of interaction between the receptor and the hydrocarbon molecule. Consequently, other strategies have been developed to overcome this shortfall. Rather than design one receptor that binds specifically to a particular analyte, a range of receptors with different degrees of binding can be arrayed into a sensor for the detection of many analytes. This approach has been inspired by nature and various groups have shown that polymer-based sensors used in an array format, which employ pattern recognition methods, can provide a higher degree of selectivity and reversibility (29, 32, 33). The responses of several individual sensors are probed under various conditions and multicomponent analysis is performed to locate a pattern and predict the signal of an unknown sample. The limitation of polymers has led some researchers to explore alternative receptors. Zeolites are another group of compounds that have also generated some interest (38, 39). This is not surprising considering that they comprise a microporous open framework structure which is accessible to certain guest molecules. Furthermore, their surface and structural properties can be easily modified (40), which makes them ideal candidates for the selective adsorption of small organic molecules. Several groups have shown that zeolites are promising materials for the detection of various volatile hydrocarbons (41-43). However, they do have a strong tendency to preferentially adsorb/absorb water relative to hydrocarbons (44, 45) and this appears to be hindering their application in environments with elevated moisture levels. Others have proposed that carbon nanotubes are suitable receptors for the adsorption/absorption of volatile organic compounds (46-52), noting that their high surface area and geometry-dependent electrical properties make them highly attractive materials. Studies by Penza and co-workers (4749, 53) revealed that the analytical performance of a sensor coated with carbon nanotubes was superior compared to a sensor without carbon nanotubes. By contrast, Parikh et al. (51) fabricated a single-walled carbon nanotube (SWCNT) vapor sensor by dip coating a poly(ethyleneterephthalate) substrate into an aqueous surfactant-carbon nanotube dispersion. However, the response was more favorable for polar compared to nonpolar molecules. Despite the improved sensitivity there are a number of factors that need to be considered when preparing sensors based on carbon nanotubes (54, 55). Furthermore, water and inorganic oxide vapors strongly interfere with sensor response, suggesting that these

and other polar compounds may need to be removed from the sample prior to hydrocarbon monitoring.

Transducers Mass/Piezoelectric. The piezoelectric sensor has come a long way since the work of Sauerbrey (56), which led to the development of an analytical device based on a mass change through the measurement of the resonance frequency. The quartz crystal microbalance (QCM) and the surface acoustic wave (SAW) are two types of mass-sensitive transducers that are commonly used for hydrocarbon detection. A number of articles have been published discussing the theoretical and practical aspects of QCM and SAW devices (57, 58), and it is not the purpose of this paper to discuss the relative merits of each one. Although, there is a general consensus that SAW devices outperform QCMs in terms of their sensitivity (34, 59, 60). However, some workers have shown that QCMs driven at the fundamental resonance frequency of 5 MHz can achieve high sensitivities under certain conditions (36, 61). Notwithstanding, a major pitfall of SAW-based devices is that they tend to be more sensitive to various physical parameters (57) and this issue, along with others (62), needs to be considered when developing mass sensors for hydrocarbons. A number of groups have demonstrated that when the quartz crystal is loaded with a polymer it can be used to detect mass changes as a function of hydrocarbon concentration (63, 64). These studies were performed in the gas phase and detection limits in the ppm range were achieved. However, Sugimoto and co-workers (65, 66) employed a diode-based radiofrequency sputtering apparatus to prepare a porous sintered polyethylene quartz resonator sensor and used it to detect volatile petroleum hydrocarbons down to ppb levels. Similarly, Potyrailo and Sivavec (20) revealed that a silicone block polyimide acoustic wave sensor responds linearly to trichloroethylene vapor over the concentration range of 100-1000 ppb. A dip-coating method was used to deposit the polymer onto both sides of the AT-cut quartz crystal, and X-ray photoelectron spectroscopy confirmed that little or no surface oxidation occurred after 3 years of sensor use. To overcome some of the pitfalls of polymer receptors the group of Shih (59, 60) immobilized fullerene and silver ion cryptands onto a SAW transducer and used it to detect saturated and unsaturated hydrocarbons. They found that the sensitivity depended on the type of metal ion incorporated in the cryptand, the structure of the cryptand, and on the cryptand loading content. However, the sensor was more sensitive toward polar-based organic compounds and a gas chromatography separation step was introduced to help improve selectivity. Others have shown that piezoelectric transducers coated with polyepichlorohydrine-zinc oxide (67) and cyclodextrin-titanium dioxide (68) offer a more robust and selective sensor for volatile hydrocarbons. It is evident that the responses of many different polymers toward hydrocarbons have been evaluated mainly in the gas phase; however, a number of other groups have shown that hydrocarbon detection can also be achieved in liquid samples. Ueyama and co-workers (69) used a polymer-coated QCM sensor to discriminate among various petroleum-based products. Although, the authors failed to report the type and the chemical composition of the polymer used in their studies. Applebee et al. (70) developed an acoustic wave sensor and demonstrated that it responds to various hydrocarbons over a wide linear concentration range. More importantly, measurements were performed in real water samples and validation studies revealed that the sensor compares well with gas chromatography analysis. Hossenlopp and coworkers (71, 72) also evaluated the response of several polymers for the detection of aromatic hydrocarbons in VOL. 41, NO. 18, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

6335

aqueous solutions over the range of 1-60 ppm. The viscoelastic contributions arising from sensor response were modeled and it was shown that acceptable agreement can be achieved between the experimental results and theory. By contrast, Lieberzeit et al. (73) employed a polyurethane receptor to distinguish between different engine oils and concluded that the sensor can be used to detect changes in the chemical properties of the oil. As long as piezoelectric sensors comply with the Sauerbrey equation the frequency changes that occur as a function of hydrocarbon concentration can be used to provide reliable quantitative information. Although, in some cases the frequency shifts cannot be interpreted simply in terms of hydrocarbon adsorption/absorption. Lucklum et al. (74) observed a decrease in frequency with increased trichloroethylene levels; however, at elevated polybutadiene coating thicknesses the frequency shifted in the positive direction with increasing analyte concentration. These changes were attributed to viscoelastic effects occurring in the polymer. Others also observed similar frequency changes on an acidcoated polysiloxane polymer as a function of solution pH (75). Consequently, these studies highlight that the response mechanism of polymer-based mass sensors are indeed complex and the fundamental processes which dictate their response need to be understood before reliable measurements can be made in environmental samples. The need to improve the analytical performance of hydrocarbon sensors in terms of their sensitivity and selectivity has been the impetus behind the pursuit for new receptors. Percival and co-workers (61) found that an ionically bound anthracene recognition element can be used to detect polycyclic aromatic hydrocarbons down to ppb levels in ethanol solutions. Despite the high sensitivity it appears that this approach works well only for aromatic-based hydrocarbons, which are able to undergo a π-π interaction mechanism. Furthermore, the sensor may only be applicable to certain aqueous samples since the receptor stability is likely to be influenced by pH and ionic salts. Calixarenes have also received a great deal of interest, noting that their structure and size can be tailored to provide a well-defined cavity for the molecule of interest. Several groups compared the analytical performance of various calixarenes and polymers for the detection of chlorinated and aromatic hydrocarbons in water (74, 76-78). These studies demonstrated that calixarenes are more sensitive relative to the polymer-based receptors. Lieberzeit et al. (79) revealed that the size of the upper rim of the calixarene molecule plays an important role on the sensing mechanism, while others (80) concluded that the lower rim has a significant influence on sensor selectivity. It is evident that calixarenes may overcome some of the limitations observed with polymer-based receptors; however, their analytical properties are highly dependent on the experimental conditions used during sensor preparation (81). In addition, the relatively strong interactions between calixarene receptors and hydrocarbons give rise to response curves that display irreversible behavior, and this can be a problem for real-time measurements. Sensors based on mass transduction (i.e., QCM, SAW) have been applied successfully for the analysis of hydrocarbons down to ppb levels. Despite some encouraging analytical results there are certain precautions that need to be considered if they are going to be used to generate meaningful and accurate data. A common challenge with these sensors is the danger of generating a response from anything that has mass. Minimizing nonspecific interactions between the sensor surface and interferences in the gas or liquid phase is highly desirable, and this issue has been partly addressed by the design of receptors that exclusively interact with the analyte of interest. Another major drawback of these sensors is that they also respond to changes in the physical 6336

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 18, 2007

properties of the liquid and to viscoelastic variations of the sensing film. Unfortunately, these challenges have not been satisfactorily addressed and are hindering their application to the detection of hydrocarbons in environmental waters. Optical. By exploiting certain energies of the electromagnetic spectrum a great deal of information can be obtained about the structure and identity of hydrocarbon molecules. These sensors rely on a light source and an optical transducer for signal measurement. The radiation is used to excite the molecules and the absorbed/emitted energy is monitored at a single frequency or over a range of frequencies. The light source is designed chiefly to interact with a reagent or receptor molecule, which has been immobilized on a solid substrate. Despite the existence of several modes of optical measurement, sensors based on infrared spectroscopy (IR) are the most common methods for hydrocarbon detection (82, 83). This is not surprising considering that IR sensors are versatile and have been used to study gas, liquid, and solid samples. Although, some propose that the fluorescence-based optical sensor may be more sensitive and suitable for the detection of aromatic hydrocarbons (84). It is well-established that every hydrocarbon molecule generates a characteristic absorption pattern or “fingerprint” and the ability to distinguish between different compounds based on the infrared absorption spectrum is an attractive feature that has been exploited by many groups. Henning et al. (85) developed a near-infrared hand-held tunable laser diode and showed that it can distinguish between methane and ethane within several seconds. Likewise, Stewart and co-workers (86) designed a portable IR sensor for methane and demonstrated that it behaves reliably in harsh environments over a wide temperature range. It is worth noting that these sensors do not comprise a receptor layer, suggesting that environmental studies are likely to be limited to simple volatile mixtures. Although, the levels of hydrocarbons typically found in the environment are so low (∼ppb) that their detection using the aforementioned IR techniques may be problematic. Consequently, various strategies have been developed over the past decade to overcome this and other challenges. Attenuated total reflectance (ATR) is one approach that has been successfully used to tackle the detection of hydrocarbons in aqueous solutions. Indeed, ATR is a surfacesensitive technique that has been used to identify and quantify various hydrocarbons adsorbed/absorbed in Teflon (87, 88), poly(dimethylsiloxane) (88), poly(styrene-co-butadiene) (88), ethylene-propylene copolymer (89-91), polystyrene (92), poly(acrylonitrile-co-butadiene) (88, 93), cyclodextrin-polyvinylbenzyl (94), and zeolites (43). These studies have demonstrated that ppb detection levels can be achieved under certain conditions, noting that the receptor also plays an important role in sensitivity. Normally, the direct detection of hydrocarbons in water using IR is difficult due to absorption by the O-H group. However, Albuquerque et al. (95) managed to overcome this by developing a poly(dimethylsiloxane) rod near-infrared optical fiber transflectance sensor and used principal component analysis to identify the various hydrocarbon mixtures. The sensor response kinetics and sensitivity were shown to be highly dependent on the geometrical properties of the polymer. Despite the promising results further work is needed to improve the limit of detection, which varied between 3 and 8 mg/L. Another problem particularly with sensors in aqueous environments is the possibility of water facilitating the degradation of the receptor and changing its response characteristics. To minimize degradation issues Yang and Ramesh (96) employed a porous Teflon membrane over the polyisobutylene receptor and were able to detect several different volatile organic compounds down to ∼250 ng/mL. Surface plasmon resonance (SPR) is another surfacesensitive technique that has also attracted some interest and

various groups have demonstrated its potential for the detection of hydrocarbons (97-99). Despite displaying linear changes in the resonance angle as a function of hydrocarbon concentration, these sensors do not appear to be very sensitive and further work is needed to evaluate their analytical performance in real samples. Many of the aforesaid optical sensors studies have been performed using large laboratory-based instruments that require careful experimental control. Consequently, there is a push for robust sensors that can be deployed directly in the field with limited operator handling. There is no doubt that the use of fiber-optic technology has increased the versatility of optical sensors and this makes them a powerful analytical tool. A number of groups have studied the response characteristics of various polymer-coated fiber optic sensors for the detection of a wide range of hydrocarbons in waters and soils (100-108). These studies have demonstrated that low hydrocarbon levels can be detected rapidly (