PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
 
Sens Actuators B Chem. Author manuscript; available in PMC 2010 July 16.
Published in final edited form as:
Sens Actuators B Chem. 2009 July 16; 140(2): 490–499.
doi:  10.1016/j.snb.2009.04.053
PMCID: PMC2765673
NIHMSID: NIHMS130967

Selective Detection of Sulfur Derivatives Using Microfabricated Tuning Fork-Based Sensors

Abstract

The paper describes an integtrated sensor system that can selectively and reversibly detect sulfur derivatives in the presence of interferent molecules. This is accomplished by integrating analyte-specific sensing materials with optimized filter materials. Microfabricated quartz tuning fork arrays are used to provide fast, accurate and low-cost transduction of the analyte binding events into electronic signals. The concept is demonstrated for detection of three sulfur derivatives – dimethyl disulfide, ethanethiol and methylsulfide.

Keywords: sulfur derivatives, thiols, disulfides, sulfides, quartz tuning fork, selective detection

1. Introduction

Many chemical sensors rely on molecular recognition, or specific binding of an analyte to a proper sensing material. Various materials, including organic molecules, conducting and non-conducting polymers and metal nanoparticles, are incorporated into sensors that are capable of transducing the binding event into a signal corresponding to an electrical, electrochemical, optical or mechanical property change [1]. An alternative approach is based on pattern recognition or electronic noses, using chem-resistors [2,3], colorimetric [4,5], surface plasmon resonance, microcantilever, surface acoustic wave and quartz crystal sensors [6]. The signals obtained from the electronic noses are analyzed with algorithms (e.g., Principal Component Analysis [1,7,8], Fisher Linear Discriminant [7,8], Hierarchical Cluster Analysis [4]) which, when combined with training using odor libraries, provide useful tools to identify specific odor patterns associated with particular tasks such as food quality control [6,7]. In spite of advances with these approaches, identifiying and detecting multiple trace analytes in a complex mixture of interferents still faces formidable challenges.

This paper presents an alternative strategy to improve selectivity and accuracy of detection of a particular chemical vapor or group of vapors, namely sulfur derivatives. In the current work, this technique has been optimized for indoor air analysis, such as inside a laboratory, or a household kitchen or bathroom. It includes: (1) a system-level approach that treats both an analyte-specific sensor and an interferent-filter with equal emphasis, (2) an efficient screening method to determine analyte-specific sensing and interferent filtering materials, and (3) a suitable transduction mechanism chosen based on the physical and chemical properties of the target analytes. We also emphasize the need of testing a sensor for its capability not only to sensitively detect a target analyte but also to reject common interferents. This concept is used to selectively detect sulfur derivatives (malodor molecules) at concentrations as low as several parts per million (ppm) in the presence of interferent molecules. A similar concept has been applied and succesfuly proved in commercial sensors and to design sensors for breath analysis [9].

Organic Sulfur Compounds (OSCs) are commonly used as products in the manufacturing of wood pulp, fuels and detergents [10]. They are also released in various commercial processes and known to play a significant role in global temperature and acid precipitation [10]. In addition, OSCs are indicators of euthrophication, environmental pollution of natural water sources [3] and bad indoor ventilation. Release of OSCs also causes odor problems, and some of them are toxic in nature. Thus, it is important to detect and quantitatively assess the concentrations of these gases in ambient environments. Thiols and sulfides are often detected using gas chromatography (GC) [10]. While GC is a powerful and sensitive tool [11], it is still an expensive and user-dependent technique for real-time continuous monitoring purposes.

Scheme 1 illustrates the sensing device including an array of sensing elements made of microfabricated Quartz Tuning Forks (QTF), a filter subsystem and sample delivery subsystem (a pump (shown) and valve) that are controlled with a microcontroller (not shown). Each QTF sensing element has a polymer wire stretched across the two prongs, and a specific binding of the analyte triggers a change in the mechanical properties of the polymer wire, which is quickly and sensitively detected via a change in the resonance frequency of the QTF. Depending on the specific nature of interactions, the frequency can either increase or decrease, providing more selective detection than mass-loading based chemical sensors (described below). The filter subsystem not only protects the sensing elements from over exposure and fouling, but also further improves the selectivity of the entire sensor system especially in presence of high concentrations of interferent molecules often found in real work applications. The microcontroller provides intelligent control and consorted operations of the sampling, filtration and detection. Surface Plasmon Resonance Imaging (SPRi) and QTF are used to screen various polymers for both the filter and sensing materials.

Scheme 1
Schematic representation of sulfur derivative sensing system. The device is an integration of sensitive sensing elements and a filter subsystem controlled with microcontrollers. Each sensing element consists of a tuning fork and a polymer wire stretched ...

2. Experimental

2.1 Reagents

Analytes, dimethyldisulfide (DMS), methylsulfide and ethanethiol, and interferents, Dowanol TPM (a major component in cleaners), Benzyl Acetate (BA, a fragrance component), ammonia, hydrochloric acid, toluene, isopropanol, acetone and acetaldehyde were obtained from Sigma-Aldrich, Co. and used without further purification during the screening process and sensor performance evaluation. Other interferents from commercial products: general purpose cleaner (3M), bathroom desinfectant cleaner (3M), wax cleaner (Butchers® Amplify, 3M), body spray (Axe), body mist (Coty, US) were employed according to the manufacturer instructions or as per routine use. The products were applied on the floor and surfaces or on a person’s body inside a small room (10 ft. × 7.5 ft. × 10 ft.). Air samples in the room, containing realistic concentrations of these products, were collected for testing. Saturating vapors of mold obtained from a bath tub were collected in a Tedlar® bag and analyzed.

Filter and sensing materials studied in this work include Ethyl Cellulose (EC) with different substitution degrees: T-10 (S.D. = 50.6%), N-22 (S.D. = 49.0%) and N-7 (S.D. = 48.5%), hydroxypropyl Cellulose (HPC) and sodium CarboxyMethyl Cellulose (CMC) purchased from Hercules Inc. Aqualon Division. Other polymer materials include Wax (Apiezon®), Polycarbonate (PC, Acros Organics), Polystyrene (PS, Acros Organics), Poly(methyl methacrylate) (PMMA, Acros Organics), Carbowax® (CW – MTO,,Supelco), Poly(vinyl chloride) (PVC, Sigma-Aldrich), Parafilm® M (PF, Sigma-Aldrich), Cytop® (Bellex International Co.), Polyaniline (emeraldine salt, PANI – Aldrich), Poly(3-hexylthiophene-2,5-diyl) (PHTP, Aldrich), Poly(ethylene glycol) (PEG – Aldrich), Poly(acrylic acid) (PAA, Aldrich), Poly(diallyldimethylammonium chloride) (PDDA – Aldrich), Poly(vinylsulfonic acid, sodium salt; PVS - Aldrich), Poly (sodium-4-styrenesulfonate, PSS –Aldrich), Methyl Cellulose (MC, Sigma-Aldrich), Polyethyleneimine (PEI, Aldrich), Mercury Chloride (Sigma-Aldrich) and Gold (Kurt J. Lesker, Clairton, PA).

2.2. Surface Plasmon Resonance imaging (SPRi)

SPRi was used to perform initial screening of materials to identify differential affinities between representative components of malodor (DMS), cleaner (Dowanol) and a fragrance (BA). The principle of SPRi is explained elsewhere [1216]. Briefly, a p-polarized collimated light beam from a 648 nm LED is directed through a prism onto a gold film (thickness ~50 nm) supported on a BK7 glass cover slide. The reflected light from the film is collected by a charge-coupled device (CCD) camera. The incident angle (θi ~ 43°) is chosen so that the light is partially absorbed by surface plasmons in the gold film. The absorption is extremely sensitive to changes in the refractive index and/or thickness of materials deposited on the gold surface, which can be detected and imaged by the CCD camera.

In order to use SPRi as a high throughput screening tool to determine interactions between various polymers and analyte/interferent molecules, low-concentration solutions (<200 μg/mL) of different polymers were manually deposited on the gold film using a microsyringe (0.1–0.6 μL), which resulted in formation of a patterned array of polymer films after evaporation of the solvent. The film thickness of each array element was estimated by comparing the measured SPR signals with those calculated using Winspall (http://www.mpip-mainz.mpg.de/knoll/soft/). Although the film thickness can be varied over a wide range, only those with thicknesses less than 20 nm were evaluated since this range ensures linearity of the SPR response.

A flow cell made of Teflon® with inlet and outlet was used in the screening experiment. The analyte and interferent vapors (150 ppmV DMS, BA and Dowanol) were introduced into the cell with a Pocket Pump® (SKC Inc., Eighty Four, PA) at a rate of 200 mL/min and allowed to interact with the polymer array on the sensor surface for 9 minutes.

We note that since the gold films in the SPRi screening and the silver electrodes on QTFs (described below) are highly reactive to DMS, both the gold films and QTF electrodes were passivated by exposing them to saturated DMS vapors at room temperature for 30 minutes followed by thorough washing with acetone before experiments.

2.3. QTF Sensor Arrays

An array of QTFs was used both as an additional tool to screen sensing and filter materials and as the sensing platform for the detection of sulfur derivatives (Scheme 1). QTFs are piezoelectric devices that can be easily driven into resonance with an electrical signal. The resonant frequency f0 is given by [17]

f0=12πkM
(1)

where k′ and M are respectively the effective spring constant and mass. The QTFs used in our experiments have a resonance frequency of 32768 Hz with dimensions: 4 × 0.35 × 0.6 mm3 for each prong (Newark Inone Electronics). The effective spring constants of these QTFs are ~20 kN/m and the thermal noise is as small as ~4 × 10−4 nm (root mean square oscillation amplitude of the prongs) at room temperature [18].

The outer packaging was removed and the QTFs were chemically modified with polymer wires stretched across their prongs using the following procedure [19]. First, a viscous solution of the polymer was prepared in DMS, and two small drops (~0.1 μL) of this solution were then placed on each of the two prongs of the tuning fork. Before the drops dried out, a syringe needle was dipped into one of the polymer solution drops and used to pull the drop to the second prong. Surface tension led to capillary thinning and formation of a stretched polymer wire (of diameter between hundreds of nm to tens of microns). This wire increased the effective spring constant of the QTF and consequently increased f0. The polymer wire-modified QTFs were left to dry overnight before use. Typically, the quality factor (Q) [20] of the QTFs decreases from 8,000–11,000 to 1,000–3,000 after the modifications. For mass loading applications, QTFs were simply coated with dilute polymer solutions (<0.35 μL).

A 0.2V AC signal from a function generator (Stanford Research Systems, SRS DS345) was used to drive an array of wire-modified QTFs into oscillations. The electrical outputs from the oscillating QTFs were amplified with a current amplifier and sent to a lock-in amplifier (Stanford Research Systems, SRS SR830). The outputs from the lock-in amplifier were recorded as a function of excitation frequency with a digital oscilloscope (LeCroy Wavepro 950) with a noise level of 20 mHz per second of data acquisition. A gas diluter (Model 1010 from Custom Sensor Solutions Inc.) provided chemical vapor mixtures at different concentrations. Various flow rates, sample injection and purging times, and analyte concentrations were used in the experiments.

For quick screening of filter materials, QFTs were used as mass loading sensors. An array of 9 QTFs coated with different polymers (Table 1) was used to identify potential filter materials based on their specific interactions with various interferents (organics, commercial and mold). The initial selection of these polymer materials was based on the results of the SPRi screening process described earlier. The materials were coated on the QTFs using a spin coater (Model 6700, Speciality Coating Systems, Inc.). The QTF array was driven into resonance with a microcontroller-based circuit which recorded frequency of each tuning fork by counting the number of oscillations in a fixed time pulse. The frequency resolution is limited by control circuit noise level and is ~5 mHz for the digital control circuit indigenously developed by us. Vapors of different interferent chemicals were drawn into a 15 mL test chamber using a pump (~650 mL/min.). After establishing a baseline with ambient air, the array was subjected to a 200-second injection for each test vapor, followed by a 10-minute purging period to allow the QTFs to recover to their original frequency.

Table 1
Polymers coated on QTF array for filter material screening

2.4. Interferent Filter Assembly

The filters were made by first coating clean glass fibers with solvent-based solutions of PVC or EC N-7. After the solvent was dried up, the coated fibers were packed into a clean glass tube. The filter was further baked in an isothermal oven at 60 °C for 30 minutes to release trapped solvent vapors.

3. Results and discussion

Our sensor system has two components that interact with chemical vapors – an adsorbing filter for a wide range of interferents and a QTF sensor array that responds to sulfur derivatives. Selecting suitable materials for both components is critical for selective and sensitive detections of analytes. We describe below the filter and sensing material screening results first and then discuss the performance of the sensor system.

3.1. Filter Material Screening

3.1.1. QTF Arrays

As we have already mentioned, a QTF can detect a change either in the effective mass due to analyte adsorption or spring constant resulting from interactions of analytes with the polymer wires. The relatioship between the measured resonance frequency shift of a QTF and changes in the effective mass and spring constant is described by [21]:

Δf(k,M)=(f02)[δkkδMM]
(2)

where Δk′ and ΔM denote the changes in the effective spring constant and mass, respectively. Our previous work [19,22,23] has determined that Δk′ is dominant when an analyte adsorbs onto a polymer wire stretched across QTF prongs, as ΔM is negligible when compared to the total mass of the QTF. We also found that detecting Δk′ can provide extra selectivity (see below).

When QTFs modified with polymer coating films are used to screen materials for filter design, eq. (2) shows that frequency change is solely due to mass loading effect (δM), since δk′ is zero in this case. The resonance frequency decreases with an increase in the mass which provides a straightfoward method to screen materials for filters. Rather than using multiple filters to target individual interferent molecules, a single filter that exhibits moderately high affinity for most interferents will make the sensor system more portable. More importantly, resistance associated with each successive filter will decrease the uniformity of air flow. A single filter using a mixture of different materials is another possible approach but was not explored in this study. The aim of the filter screening process is to find a single material for use as a general filter for the interferents involving organics, commercial products and mold.

The QTF array in Table 1 was exposed to analyte and interferent vapors, and the response of the array is summarized in Fig. 1. The frequency change of each QTF in the array is normalized by the frequency shift due to the coating mass of the polymer. A bare QTF was used as control. It was observed that all three ethyl cellulose polymers (EC N-22, EC N-7 and EC T-10) recorded highest changes in frequency when exposed to chemicals containing BA derivatives (spray, mist) and Dowanol derivatives (cleaners and disinfectants). Of the EC polymers, EC N-7 demonstrated least affinity for DMS. From the combined SPRi and QTF screening results, two materials were selected for further testing – PVC to filter out BA and Dowanol specifically (see below), and EC N-7 as a general filter.

Fig. 1
Response of a QTF array to DMS and interferent molecules. Each QTF is coated with a separate material (listed along X-axis). Response is normalized by frequency shift due to mass of coating material.

3.1.2. SPRi

As discussed before, SPRi-based screening was carried out by patterning polymers on the gold film and exposing them to vapors of the analyte, DMS, and interferent molecules at known concentrations. Figures 2(a)–(c) show the SPRi responses of a passivated gold film (which acts as a control) and two polymer-modified gold films (EC N-7, EC T-10) to vapors of DMS, Dowanol and BA (150 ppmV). In the control experiment (Fig. 2(a)), the three chemicals caused similar SPR responses. In the case of polymer modified gold films (Fig. 2(b)–(c)), the responses to analyte and interferent molecules are different due to each material’s unique affinity to the chemical vapors. This interaction was determined from the SPRi responses (ΔRp, the change in optical reflectance), which is related to the surface coverage of adsorbed molecules (Γ) [12,24].

Fig. 2
SPRi response (Reflectance change) to 150 ppmV vapors of analyte (DMS) and interferents (Dowanol, BA). (a) Response of passivated gold film (control). (b) Response of gold film patterned with EC-N7. (c) Response of gold film patterned with EC T-10. Pattern ...

Figs. 2(b)–(c) show that (1) for a given polymer, SPRi responses normalized by the polymer thicknesses are constant and reproducible. This is expected because the polymer films are thin (<20 nm) as compared to the decay length (~200 nm) of the evanescent field, and the SPRi reponse is linearly proportional to the adsorbed materials. Fig. 2(d) is a comparison of relative reflectance changes (ΔRp,i/ΔRp,min. = Γimin.) of DMS, Dowanol and BA on different polymers. To simplify data interpretation, the reflectance change of the molecule with lowest response (ΔRp,min.) was taken as unity and the ΔRp/ΔRp,min. ratios were rounded off. Hydrophilic polymers (PSS, PVS, PEI, PEG and MC) did not show affinity for any of the tested molecules and are excluded from Fig. 2(d). Several conclusions can be drawn from the SPRi screening results:

  1. PVC absorbs both interferents (Dowanol, BA) but does not interact with DMS. Thus, it is a good candidate for the filter material;
  2. EC T-10 and N-7, PF, CW and a mixture of PMMA/PHTP show high affinity for BA vapors;
  3. PAA is a good sorbent for selective adsorption of Dowanol;
  4. PS, Wax, EC N-22 and PHTP can be used as sensing materials to detect DMS when used in combination with filter materials mentioned above (1–3).

It was also observed that DMS has faster adsorption kinetics than Dowanol and BA, on PHTP, at a similar concentration (not shown). This is important especially if this material is to be used in differential adsorption kinetics detection methods.

It should be noted that DMS showed, as expected, a strong irreversible binding on bare gold surface, while BA and Dowanol did not. Although this material could be a choice for selective detection of sulfur derivatives against interferents, it was excluded because of the irreversible response (see section 3.5. Real sample analysis).

3.2. Sensor Material Screening for Detection of Sulfur Derivatives

After identifing the optimal materials for the filter subsystem, focus was shifted to studying the interactions of the materials with the analyte vapors. As mentioned before, sulfur derivatives are responsible for indoor and outdoor malodors, and many of them are toxic in nature [25,26]. Assessment of their levels is an important step towards providing healthy indoor and outdoor environments. This section describes the effects of DMS and some common indoor interferents on the mechanical properties of modified QTF sensors. QTFs are very sensitive and inexpensive resonators that can be modified for sensing applications [19,22,23,2729]. By monitoring changes in the resonance frequencies of modified QTFs due to chemical vapors, it is possible to study the effects of these vapors on the effective spring constants of the QTFs [19,22].

QTF arrays were modified with materials that had been classified as sensitive to DMS during the SPR screening process (previous section). Fig. 3(a) illustrates the frequency spectrum of a QTF array consisting of 6 sensing elements: EC N-22 and PHTP-coated QTFs, wax-, PS1- and PS2-wire modified QTFs, and a bare control QTF. Each peak in the spectrum is the response of one sensing element. Fig. 3(b) shows the response of the QTF array to DMS, Dowanol (390 ppmV each) and BA (150 ppmV). Note that both Dowanol and BA were saturated vapors in air. The Q-factor of the wax-polymer modified QTF was too small (broad peak in Fig. 3(a)) for accurate measurements. Significant positive frequency shifts were observed for the PS-wire modified QTFs in the presence of DMS, indicating that the analyte increased the effective spring constant of the PS wire. This is interesting because most analytes soften the polymer wires and decrease their spring constants, leading to negative frequency shifts. The positive frequency shift was confirmed in another experiment, in which a PS-wire QTF was excited at a frequency fixed near the resonance peak and the change in the amplitude due to the shift in the resonance frequency was monitored (Fig. 3(d)). As shown in Fig. 3(c)), the resonance frequency shift is linearly proportional to the amplitude change within a certain frequency window (marked between dashed lines).

Fig. 3
(a) Frequency spectrum of a QTF array including 6 sensing elements: EC N-22 and PHTP-coated QTFs, wax-, PS1- and PS2-wire modified QTFs, and a bare control QTF. PS1 corresponds to QTF modified with 2 wires. (b) Resonance frequency shift pattern for each ...

3.3. Integrated operation of sensing and filter subsystems

So far, all experiments in this work had been directly concerned with identification of suitable materials for analyte sensing and interferents’ adsorption. Having selected PS as a sensing material and PVC, EC N-7 for filtering, the filters were characterized by performing detection of DMS in the presence of saturated vapors of interferent molecules, including Dowanol and BA as well as realistic concentrations of general interferents. A filter was placed in the path of air flow before the sensing chamber (shown in Scheme 1) with the intention of trapping interferent molecules and allowing only the analyte to pass through.

Figs. 4(a)–(b) show the responses of a PS-wire modified QTF to saturated vapors of Dowanol and BA, both with and without a PVC filter in place. Frequency shift for a particular QTF is measured relative to a baseline, which is obtained by placing a zeroing filter in the air flow path to the sensor. This zeroing filter (made of Purafil) is used to obtain a baseline that is free of sulfur derivatives and other interferents, and reflects the drift of the QTF over time. Typical drift signals were several times smaller (~10) than the response signals assessed from low ppmV levels of sulfur derivatives within 10 minutes of testing (sampling and purging) time. The data in Fig. 4 clearly demonstrates the effectiveness of PVC as the filter material and the finding is consistent with the reported study of PVC interactions with ether-bond molecules [30]. Fig. 4(c) shows overall response of the sensor to DMS and to saturating vapors of Dowanol and BA, which further demonstrates selective detection of DMS.

Fig. 4
(a)–(b) Responses of a PS-wire QTF to saturated vapors of BA and Dowanol with and without a PVC filter. (c) Selective detection of DMS with the PVC filter in the presence of BA and Dowanol vapors.

The filter screening data also show that EC N-7 interacts with most interferents studied here. EC N-7 was characterized as a filter material in two cases. The first case applied to interferents whose concentrations could be accurately determined. This included single chemical component interferents, such as isopropanol (ISP) and toluene. The second case involved interferents that were either a complex chemical mixture (like cleaners, which contain Dowanol; or spray and mist, which contain BA) or chemicals with high vapor pressures, such as acetaldehyde. Figs. 5(a–b) compare the responses of the PS-wire QTF sensor to 148 ppmV of toluene and acetone, with and without an EC N-7 filter. Results of the PS-wire QTF sensor performance and EC-N7 filter with analyte (DMS) and interferents present are highlighted in Fig. 5(c). In agreement with the screening data, the EC N-7 filter absorbs interferents to a large extent while allowing DMS to pass through to the sensor elements.

Fig. 5
Response of a PS-wire QTF sensor with and without an EC N-7 filter to 148 ppmV of (a) toluene and (b) acetone (c) Response of the sensor with the N-7 filter to 148 ppmV of analyte (DMS) in the presence of interferents at same concentration.

While EC N-7 and PVC proved to be good for filtering out organic interferents (such as acetone and toluene or Dowanol and BA, respectively), their affinities for inorganic vapors like hydrochloric acid (HCl) were limited (not shown). Instead, filter materials based on their acidic/basic natures were found to be better choices for the inorganic vapors. The response of the PS-wire QTF sensor to HCl, NH3 and NaClO (from bleach) at 148 ppmV concentration in the absence of filters is shown in Fig. 6(a). It was observed that the effects of NH3 and NaClO on the PS-wire QTF were insignificant. On the other hand, HCl induced a negative frequency change of relatively high magnitude, which is indicative of polymer wire softening. Fig. 6(b) demonstrates the effectiveness of pH paper as a filter for HCl. As with the polymer materials used to filter out organic and commercial products, the pH paper filters were re-usable, with QTF sensor frequencies recovering to their original values after the system being purged with clean air. Table 2 summarizes the performance of the filters. For each interferent, the frequency change from interferent vapors relative to the frequency change due to DMS is used to measure effectiveness of the filter for that interferent. All three filters (PVC, EC N-7 and pH paper), placed before the sensing chamber, caused this ratio to decrease for each interferent introduced into the system. This substantiated the choice of these materials as filter materials. It should be noted that filter characterization experiments were carried out in environments where relative humidity variations were as high as 35%.

Fig. 6
(a) Response of a PS-wire QTF sensor to HCl, ammonia (NH3) and sodium hypochlorite (NaClO, bleach). Neither NH3 nor NaClO shows significant adsorption on PS-wire. (b) Response of the sensor to HCl vapor, using pH paper filters. Filter 1 (0.132 g) is sufficient ...
Table 2
Percentage frequency change (ΔfInterferent/ΔfDMS×100) values for the interferents listed in Scheme 1, both in the absence and presence of filters

3.4. Analytical Characterization of the Sensor

After optimizing the filtration system and sensing platform, it was necessary to determine the reproducibility, stability, detection limit, dynamic range and specificity of the entire system.

3.4.1. Reproducibility

The responses of the PS-wire QTF sensor to four consecutive injections of DMS (390 ppmV) on the same day and under nearly identical conditions of flow rate and relative humidity were found to be highly reproducible with a relative error of ~7%. A similar level of reproducibility was observed between responses to DMS (390 ppmV) on two consecutive days (not shown). Importantly, the responses of two different PS-wire QTF sensors to DMS were found to within 15% of each other. The kinetics of the adsorption – desorption process of DMS on PS is [22]:

DMS+PSkdkaDMSPS
(3)

ka and kd are the adsorption and desorption kinetics constants, respectively. Along with dissociation constant KD, their values were estimated for future injection and purging times. The results were: ka = 3.0 (SD = 1) × 10−5 s−1 ppmV−1, kd = 1.4 (SD = 0.2) × 10−3 s−1 and KD = 47 (SD = 10) ppmV. Relative errors were less than 14% and 20% for kinetics parameters and thermodynamic constant, respectively.

3.4.2. Reusability, Calibration and Dynamic Range

In order to test for reusability of the PS-wire QTF, the sensor was exposed to fifty consecutive injections of 24 ppmV DMS after exposure to a single injection at high concentration (394 ppmV), and the response was found to within 10% (Fig. 7(a), inset), demonstrating good reversibility after intense use.

Fig. 7
PS-wire QTF sensor characterization. (a) Stability of PS-wire QTF over 50 cycles of 40-sec. injections of DMS (24 ppmV) after an initial injection of 394 ppmV DMS followed by recovery. Statistical data for the 40-sec. injections included in inset. (b) ...

The PS-wire QTF was tested as a function of analyte concentration (Fig. 7(b)–(c)). The frequency change of the sensor was linearly proportional to the logarithm of DMS concentration, which is consistent with the Langmuir-like adsorption features of the analyte molecules on the sensing polymer. The reproducibility of sensitivity (defined as the slope of the semi-logarithmic plot) was found to be 11% (for a flow rate of 0.7 L/min.) and 7% (0.3 L/min.) in three different runs for each flow rate.

3.4.3. Detection Limits

Detection limit was calculated as the concentration of the analyte that caused a frequency shift in the sensor equal to three times the signal noise level. Without a filter, minimum detection limit of PS-wire QTF sensor for DMS was found to be 750 ppbV for 200-second injections at 0.66L/min. When a PVC filter was used, minimum detection limit for DMS in the presence of saturating vapors of interferents (390 ppmV Dowanol and 150 ppmV BA) was estimated to be 4.6 ppmV. In this case, injections were of 150-second duration at 0.25 L/min.

3.4.4. Sensor Lifetime and Stability

Besides good analytical performance, highly reproducible and reversible response as shown in previous sections, the PS-wire QTF sensors showed high robustness with long-term use and environmental stress. The sensors maintained its quality factor within a 16 % variation over a testing period of 15 months under hundreds of cycles of use and diverse ambient conditions (Fig. 8a). Along this line, a high immunity towards humidity was also observed with changes in quality factors and intrinsic sensitivities (amplitude/frequency slope) smaller than 1% between 10–100 % relative humidity (Fig. 8b).

Fig. 8
(a) Spectrum of a TF sensor before and after 15 months of intensive use in ambient conditions. Only a 16% change of the TF quality factor (Q) was observed. (b) Percentage of sensitivity change of a TF upon 10 – 100 % RH exposure. (c) Response ...

3.4.5. Specificity to Sulfur Derivatives

The PS-wire QTF sensor is sensitive to sulfur derivatives. Response of the sensor was determined for two sulfur derivatives (other than DMS): methylsulfide and ethanethiol. Exact concentrations of these compounds were difficult to determine due to their volatile natures. However, it was observed that the PS-wire QTF responded in a similar manner to all three analytes, showing that they all have a similar effect on the PS-wire (i.e. increasing its effective spring constant k′). Although the EC N-7 filter showed a relatively higher affinity for ethanethiol and methylsulfide than for DMS, the sensor exhibited a clear response to all three analytes even with the filter present. Table 3 summarizes response of the sensor system to all 3 sulfur derivatives. We noted that in the case of ethanethiol, absorption of analyte by sensor is as high as 48 %. However, this sensor system is intended to be a semi-quantitative tool that detects the presence of sulfur compounds in an indoor environment (see section 3.5 Real sample testing).

Table 3
Response of PS-wire QTF to sulfur derivatives

3.5. Real sample testing

PS-wire QTF sensors were exposed to malodor samples to test their capability to identify the presence of sulfur-containing components. From a series of commercial malodor samples (Robertet Fragrances, Inc.; bathroom, kitchen and pet), the PS-wire QTF sensor could positively detect the presence of sulfur compounds from pet malodor, matching the qualitative diagnosis performed by human testers who could discriminate the characteristic odor of sulfur derivatives.

The sulfur-compound positive sample was also utilized to compare the recovery capability of our proposed PS-wire based sensor to another sensor based on gold nanoparticles, a typical well-known sulfur binding material. A sensor array based on PS-wire QTF, gold nanoparticle (Au NP)-QTF and a bare QTF was built and tested for this purpose. Fig. 8c shows the response of this array towards three successive injections of the pet malodor separated by intermitent purging periods with ambient air. Despite the fact that Au NP-QTF produced a high response, a systematic decrease of the response associated with a cumulative quasi-reversible recovery was observed. On the contrary, the response of the PS-wire QTF sensor showed full recovery, confirming the advantage of using a optimized reversible binding chemistry to produce long lasting sensors for sulfur derivative detection. Furthermore, sulfur derivative concentrations in a real malodor sample with high background of glycols were estimated from our PS-wire QTF sensor and showed good agreement with total sulfur derivative concentrations assessed from GC/MS tests.

4. Conclusions

This work has demonstrated a system-level approach for designing a sensor system that exhibits reversible, sensitive and selective detection of sulfur derivatives in the presence of saturating and realistic concentration of interferents. Specific interactions of analyte molecules with polymer wires cause changes in the mechanical properties of the wires which are amplified and detected by microfabricated tuning fork sensors. Proper filter materials are used to protect the sensing elements and to further improve the selectivity of the entire system which provides robust and reversible analyte detection. The integration of the sensor and filter components and the equal emphasis on the two components are critical for the success of the approach. The work underscores the necessity of testing interferent species while developing sensitive sensing elements if one wishes to detect trace analytes in complex environments. The device consists of low cost and miniaturized components which is ideal for many indoor applications.

Acknowledgments

The authors thank K. Foley for useful discussions about Surface Plasmon Resonance imaging. This work was supported by NIH (U01 ES0106064-01). The contents of this work are solely the responsibility of the authors and do not necessarily represent the official views of NIH.

Biographies

• 

Anant Rai received his BE degree in Instrumentation and Control Engineering from the University of Delhi (India) in 2007 and is currently pursuing his PhD degree in Electrical Engineering at Arizona State University in Tempe, AZ. His main research interests include development and integration of bio/chemical sensors for biomedical applications.

• 

Francis Tsow is an Assistant Research Scientist in the Department of Electrical Engineering at Arizona State University in Tempe, AZ. He joined ASU in 2008 as a Postdoctoral researcher in Dr. Nongjian Tao’s group after receiving his PhD in Electrical Engineering in 2008 from the same group. His current research interests include integration of chemical, health care, and defense related sensors.

• 

Jeffrey Bankers received his B.A. in Chemistry at Saint John’s University in Collegeville, MN in 1997 and his Ph.D. in Materials Chemistry at the University of Minnesota, Twin Cities in 2004. He is currently a Principal Scientist working in the Home Care group at the Dial Corporation. He currently holds 4 U.S. patents with 7 more pending. His current research interests are VOC sensors & malodor reduction compounds.

• 

Erica S. Forzani is an Assistant Research Professor in the Department of Electrical Engineering at ASU in Tempe, AZ. She joined ASU in 2003 as Research Associate in NJ Tao’s group after receiving her PhD in Chemistry in 1999 from Cordoba National University, Argentina and a postdoctoral degree in University of Buenos Aires, Argentina (2000–2003). Her current research interest is in the development and integration of sensors for environmental and health care applications.

• 

Nongjian Tao joined the ASU Faculty as a Professor of Electrical Engineering and an Affiliated Professor of Chemistry and Biochemistry in August 2001. Before that, he worked as an Assistant and Associate Professor at Florida International University. In 2008, he was appointed as Director of the Center for Bioelectronics and Biosensors at the Biodesign Institute at ASU. His research interests focus on chemistry, physics and engineering of sensor integration for applications in real word environments as well as molecular electronics, nanostructured materials and devices, and interfaces between biological molecules and solid materials.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References

1. Severin EJ, Doleman BJ, Lewis NS. An investigation of the concentration dependence and response to analyte mixtures of carbon black/insulating organic polymer composite vapor detectors. Analytical Chemistry. 2000;72:658–68. [PubMed]
2. Zhang T, Nix MB, Yoo BY, Deshusses MA, Myung NV. Electrochemically functionalized single-walled carbon nanotube gas sensor. Electroanalysis. 2006;18:1153–58.
3. Bouwman AF, Van Vuuren DP, Derwent RG, Posch M. A global analysis of acidification and eutrophication of terrestrial ecosystems. Water Air And Soil Pollution. 2002;141:349–82.
4. Janzen MC, Ponder JB, Bailey DP, Ingison CK, Suslick KS. Colorimetric sensor Arrays for volatile organic compounds. Analytical Chemistry. 2006;78:3591–600. [PubMed]
5. Johnson-White B, Zeinali M, Shaffer KM, Patterson CH, Charles PT, Markowitz MA. Detection of organics using porphyrin embedded nanoporous organosilicas. Biosensors & Bioelectronics. 2007;22:1154–62. [PubMed]
6. Albert KJ, Lewis NS, Schauer CL, Sotzing GA, Stitzel SE, Vaid TP, Walt DR. Cross-reactive chemical sensor arrays. Chemical Reviews. 2000;100:2595–626. [PubMed]
7. Vaid TP, Burl MC, Lewis NS. Comparison of the performance of different discriminant algorithms in analyte discrimination tasks using an array of carbon black-polymer composite vapor detectors. Analytical Chemistry. 2001;73:321–31. [PubMed]
8. Wu W, Mallet Y, Walczak B, Penninckx W, Massart DL, Heuerding S, Erni F. Comparison of regularized discriminant analysis, linear discriminant analysis and quadratic discriminant analysis, applied to NIR data. Analytica Chimica Acta. 1996;329:257–65.
9. Fernandes DLA, Rolo TA, Oliveira JABP, Gomes MTSR. A new analytical system based on an acoustic wave sensor for halitosis evaluation. Sensors and Actuators B. 2008 on line.
10. Beiner K, Popp P, Wennrich R. Selective enrichment of sulfides, thiols and methylthiophosphates from water samples on metal-loaded cation-exchange materials for gas chromatographic analysis. Journal Of Chromatography A. 2002;968:171–76. [PubMed]
11. Zhong QY, Steinecker WH, Zellers ET. Meso-scale VOC-mixture analyzer with tunable separation and microsensor array detection. Rare Metal Materials And Engineering. 2006;35:61–64.
12. Campbell CT, Kim G. SPR microscopy and its applications to high-throughput analyses of biomolecular binding events and their kinetics. Biomaterials. 2007;28:2380–92. [PubMed]
13. Chinowsky TM, Grow MS, Johnston KS, Nelson K, Edwards T, Fu E, Yager P. Compact, high performance surface plasmon resonance imaging system. Biosensors & Bioelectronics. 2007;22:2208–15. [PMC free article] [PubMed]
14. Li Y, Lee HJ, Corn RM. Detection of protein biomarkers using RNA aptamer microarrays and enzymatically amplified surface plasmon resonance imaging. Analytical Chemistry. 2007;79:1082–88. [PMC free article] [PubMed]
15. Piliarik M, Vaisocherova H, Homola J. A new surface plasmon resonance sensor for high-throughput screening applications. Biosensors & Bioelectronics. 2005;20:2104–10. [PubMed]
16. Ly N, Foley K, Tao NJ. Integrated label-free protein detection and separation in real time using confined surface plasmon resonance imaging. Analytical Chemistry. 2007;79:2546–51. [PubMed]
17. Ziegler C. Nanotechnologies for the biosciences. Analytical And Bioanalytical Chemistry. 2004;379:903–03.
18. Grober RD, Acimovic J, Schuck J, Hessman D, Kindlemann PJ, Hespanha J, Morse AS, Karrai K, Tiemann I, Manus S. Fundamental limits to force detection using quartz tuning forks. Review Of Scientific Instruments. 2000;71:2776–80.
19. Ren MH, Forzani ES, Tao NJ. Chemical sensor based on microfabricated wristwatch tuning forks. Analytical Chemistry. 2005;77:2700–07. [PubMed]
20. Friedt JM, Carry E. Introduction to the quartz tuning fork. American Journal Of Physics. 2007;75:415–22.
21. Hansen KM, Thundat T. Microcantilever biosensors. Methods. 2005;37:57–64. [PubMed]
22. Boussaad S, Tao NJ. Polymer wire chemical sensor using a microfabricated tuning fork. Nano Letters. 2003;3:1173–76.
23. Tsow F, Tao N. Microfabricated tuning fork temperature and infrared sensor. Applied Physics Letters. 2007;90
24. Shumaker-Parry JS, Aebersold R, Campbell CT. Parallel, quantitative measurement of protein binding to a 120-element double-stranded DNA array in real time using surface plasmon resonance microscopy. Analytical Chemistry. 2004;76:2071–82. [PubMed]
25. Kim KH, Jeon EC, Choi YJ, Koo YS. The emission characteristics and the related malodor intensities of gaseous reduced sulfur compounds (RSC) in a large industrial complex. Atmospheric Environment. 2006;40:4478–90.
26. Segala C, Poizeau D, Mace JM. Odors and health: a descriptive epidemiological study around a wastewater treatment plant. Revue D Epidemiologie Et De Sante Publique. 2003;51:201–14. [PubMed]
27. Zhang J, O’Shea S. Tuning forks as micromechanical mass sensitive sensors for bio- or liquid detection. Sensors And Actuators B-Chemical. 2003;94:65–72.
28. Zhou XF, Jiang T, Zhang J, Wang XH, Zhu ZQ. Humidity sensor based on quartz tuning fork coated with sol-gel-derived nanocrystalline zinc oxide thin film. Sensors And Actuators B-Chemical. 2007;123:299–305.
29. Guan SH. Frequency encoding of resonant mass sensors for chemical vapor detection. Analytical Chemistry. 2003;75:4551–57. [PubMed]
30. Nefso EK, Burns SE. Comparison of the equilibrium sorption of five organic compounds to HDPE, PP, and PVC geomembranes. Geotextiles And Geomembranes. 2007;25:360–65.