Article pubs.acs.org/Langmuir
Mercury Sorption and Desorption on Gold: A Comparative Analysis of Surface Acoustic Wave and Quartz Crystal Microbalance-Based Sensors K. M. Mohibul Kabir,† Ylias M. Sabri,† Ahmad Esmaielzadeh Kandjani,† Glenn I. Matthews,‡ Matthew Field,§ Lathe A. Jones,† Ayman Nafady,∥,⊥ Samuel J. Ippolito,*,†,‡ and Suresh K Bhargava*,† †
Centre for Advanced Materials & Industrial Chemistry (CAMIC), School of Applied Sciences, ‡School of Electrical and Computer Engineering, and §RMIT Microscopy and Microanalysis Facility (RMMF), RMIT University, Melbourne, VIC 3001, Australia ∥ Department of Chemistry, Faculty of Science, Sohag University, Sohag, Egypt ⊥ Department of Chemistry, College of Science, King Saud University, Riyadh, Saudi Arabia ABSTRACT: Microelectromechanical sensors based on surface acoustic wave (SAW) and quartz crystal microbalance (QCM) transducers possess substantial potential as online elemental mercury (Hg0) vapor detectors in industrial stack effluents. In this study, a comparison of SAW- and QCM-based sensors is performed for the detection of low concentrations of Hg0 vapor (ranging from 24 to 365 ppbv). Experimental measurements and finite element method (FEM) simulations allow the comparison of these sensors with regard to their sensitivity, sorption and desorption characteristics, and response time following Hg0 vapor exposure at various operating temperatures ranging from 35 to 75 °C. Both of the sensors were fabricated on quartz substrates (ST and AT cut quartz for SAW and QCM devices, respectively) and employed thin gold (Au) layers as the electrodes. The SAW-based sensor exhibited up to ∼111 and ∼39 times higher response magnitudes than did the QCM-based sensor at 35 and 55 °C, respectively, when exposed to Hg0 vapor concentrations ranging from 24 to 365 ppbv. The Hg0 sorption and desorption calibration curves of both sensors were found to fit well with the Langmuir extension isotherm at different operating temperatures. Furthermore, the Hg0 sorption and desorption rate demonstrated by the SAW-based sensor was found to decrease as the operating temperature increased, while the opposite trend was observed for the QCM-based sensor. However, the SAW-based sensor reached the maximum Hg0 sorption rate faster than the QCM-based sensor regardless of operating temperature, whereas both sensors showed similar response times (t90) at various temperatures. Additionally, the sorption rate data was utilized in this study in order to obtain a faster response time from the sensor upon exposure to Hg0 vapor. Furthermore, comparative analysis of the developed sensors’ selectivity showed that the SAW-based sensor had a higher overall selectivity (90%) than did the QCM counterpart (84%) while Hg0 vapor was measured in the presence of ammonia (NH3), humidity, and a number of volatile organic compounds at the chosen operating temperature of 55 °C.
1. INTRODUCTION Mercury (Hg) exposure has been considered to be one of the major global concerns for decades because of its toxicity and adverse effects of its related compounds on environmental and human health. 1−3 The growth of industrialization has significantly increased the amount of Hg in the environment, with recent reports showing that 30% of Hg present in the atmosphere originates from anthropogenic sources such as refineries, mining, coal-fired power plants, and related industrial activity.4,5 Most of the Hg (∼90%) emitted from these anthropogenic sources is in gaseous elemental (Hg0) form. To maintain mercury emission at minimal levels and comply with more stringent policies set by regulating bodies such as the World Health Organization (WHO) and United States Environmental Protection Agency (US-EPA), Hg removal © XXXX American Chemical Society
processes are being implemented in numerous industrial sources. Reliable, continuous monitoring of elemental mercury (Hg0) vapor is a key step in these efforts, as it assesses the efficiency of the implemented Hg removal technologies. Most of the current Hg0 vapor detection systems are based on spectroscopic methods such as cold vapor atomic absorption spectrometry (CVAAS) and atomic fluorescence spectrometry (AFS).6,7 These methods are capable of detecting very low concentrations of Hg0 vapor and are approved by the U.S. EPA. However, these methods are not suitable for many industrial applications, as they are not highly selective toward Hg0 vapor Received: May 21, 2015 Revised: June 25, 2015
A
DOI: 10.1021/acs.langmuir.5b01858 Langmuir XXXX, XXX, XXX−XXX
Article
Langmuir
Figure 1. (a) SAW and (b) QCM device structures considered in the FEM simulations.
diffusion studies following Hg0 vapor exposure were investigated.
when other common industrial gas species are present, posing as interfering gases.8,9 Moreover, they possess some other major drawbacks such as the requirement of sample pretreatment and instrumentation maintenance which make them unsuitable for the online monitoring of Hg0 vapor within many industrial processes.10,11 Commercially available solid-state portable mercury detectors such as the Jerome 411X (Arizona Instruments) are also being used for Hg0 vapor measurement in the workplace; however, these systems have been found to possess cross sensitivity when common interferants such as ammonia (NH3) and humidity are present.10 Methods based on gold (Au) layers such as piezoelectric microcantilevers, acoustic microsensors, surface plasmon resonance (SPR), and conductometric sensors are being investigated extensively.12−14 However, most of these methods are still laboratory-based and need to be studied further and understood before they can be reliably employed in industrial stack effluents. Mass-based transducers based on Au−Hg amalgamation are another group of Hg0 vapor sensors which have recently been shown to have substantial potential to be implemented as online Hg0 vapor detectors because of their robust nature, high sensitivity, high portability, and low cost. Our group has recently reported surface acoustic wave (SAW)- and quartz crystal microbalance (QCM)-based sensors that can be used to detect low concentrations of Hg0 vapor efficiently (as low as 24 ppbv) in the presence of common industry-related gas species such as NH3, acetaldehyde (MeCHO), and humidity and volatile organic compounds (VOCs) such as ethylmercaptan (EM), dimethyl disulfide (DMDS), and methyl ethyl ketone (MEK).15,16 It was shown that these sensors are capable of working in a range of operating temperatures and can withstand harsh industrial environments. As a consequence of their potential for use as online sensors under real world conditions, it is of high importance to study the physical interaction between Hg0 vapor and Au films for both SAW- and QCMbased sensors both theoretically and experimentally. Boundarycondition-based modeling methods such as finite element method (FEM) simulation can be implemented to study the behavior of SAW- and QCM-based sensors toward Hg0 vapor, which can further be used to optimize the transducer design. An experimental comparison of these transducers also needs further study to understand behavioral differences under a range of potential operating conditions. In this study, we have performed detailed experimental investigations and FEM modeling in order to compare SAWand QCM-based sensors for the detection of low concentrations of Hg0 vapor. The electrodes of both sensors were based on Au thin films. The sensors’ sensitivity was tested with respect to low concentrations (as low as 24 ppbv) of Hg0 vapor at a range of operating temperatures. The sensitivity, selectivity, sorption and desorption characteristics, response time, and
2. THEORETICAL MODELING A SAW-based sensor operates by utilizing the propagation of an acoustic wave on the surface of a piezoelectric material.17−19 Typically, a SAW-based sensor consists of two sets of electrodes known as input and output interdigital transducers (IDTs). An acoustic wave is generated and starts propagating along the surface of the material once an alternating voltage is applied to the input IDT. The mechanical wave can be converted back to an electrical form when it reaches the output IDT. The velocity of this propagating acoustic wave changes when perturbations occur on the surface of the electrodes because of interactions with the measured gas species. This change in the wave velocity can be interpreted as a resonance frequency (f 0) shift of the sensor. The shift in the resonance frequency (Δf) can be related to the change in Au film properties using a simple perturbation formula20 (eq 1). Δf = (k1 + k 2)f0 2 ρh
(1)
In this equation, k1 and k2 represent substrate material constants and ρ and h are the Au film density and thickness, respectively. As opposed to a SAW-based sensor, a QCM-based sensor operates by utilizing a bulk acoustic wave which is generated by applying a varying voltage to the electrode that is patterned on a piezoelectric substrate (i.e., AT-cut quartz).21−23 Typically there are two electrodes patterned on the either side of the piezoelectric substrate. The resonance frequency (f 0) of the sensor is dependent on the thickness and crystallographic properties of the piezoelectric substrate on which it is fabricated. One or both of the electrode surfaces can be made of a material which is selective to the analyte being measured (i.e., Au in the case of Hg0 vapor detection). Mass loading of the targeted gas species on the electrode surface shifts the resonance frequency of the sensor, which can be related to the concentration of the gas which is being measured by the Sauerbrey equation24 (eq 2) Δf = −
2f0 2 A ρμ
Δm (2)
In this equation, Δf represents the change in the resonance frequency, Δm is the change in mass at the surface, f 0 represents the center frequency of the QCM, A is the active area of the QCM electrodes, and ρ and μ are the crystal density and shear modulus of the piezoelectric crystal, respectively. SAW-based sensors possess higher sensitivity toward gas analytes than QCM-based sensors as more energy is confined on the surface of the device rather than in the bulk. Another B
DOI: 10.1021/acs.langmuir.5b01858 Langmuir XXXX, XXX, XXX−XXX
Article
Langmuir
Figure 2. Surface plots of the piezoelectric displacement of the (a) simulated SAW device and (b) simulated QCM device with clean Au electrodes and simulated (c) SAW- and (d) QCM-based sensors with Au electrodes perturbed by ∼0.025 monolayer of Hg. Shift in the resonance frequency of the (e) SAW- and (f) QCM-based sensors when the Au electrodes were perturbed by the Hg0 vapor concentration of 113 ppbv (equivalent to ∼0.025 monolayer of Hg on the Au surface).
main reason for the superior sensitivity of the SAW over QCMbased sensors is their higher operating frequencies (40−200 MHz compared to 5−20 MHz).21,25 To understand the relative sensitivity of these two sensing platforms toward different concentrations of Hg0 vapor, a series of two-dimensional (2D) FEM simulations were performed. Two simplified structures of SAW- and QCM-based devices were considered for the simulation. As shown in Figure 1a,b, both of the simulated structures were based on suitable cuts of the quartz substrate (ST and AT cut for SAW and QCM, respectively), and Au was used for the electrodes. The material constants used for the simulations can be found elsewhere.26 The simulated SAW structure contained five pairs of electrodes in both input and output IDTs, with each electrode having a thickness and width of 0.1 and 6 μm, respectively. The length and depth of the whole SAW device structure were 1000 and 500 μm, respectively. The simulated QCM structure contained two Au electrodes, each having a similar thickness to the SAW device electrodes. The width of the QCM electrodes was set to 240 μm with a total surface area ratio of 1:4 between SAW- and QCM-based devices. Frequency domain analysis was performed for both the simulated SAW- and QCM-based structures. In the case of the SAW structure, 1 and 0 V were applied to the odd and even input coupled electrodes, respectively, while both of the odd and even output electrode voltages were set to zero and coupled separately. For the QCM structure, one of the electrodes was set to 1 V while the other one was set to zero. Resonance frequencies of both simulated devices were determined following each simulation. The FEM mesh was arranged such that the SAW structure contained a minimum of
1 node/μm along the surface (direction of propagation). However, a relatively coarser node density was maintained into the depth of the SAW structure as it did not affect the simulation result significantly and also allowed us to maintain the total number of elements at an optimum processing level. In the case of the QCM structure, 1 node/6 μm was maintained throughout the depth of the substrate while a relatively coarser node density was used along the surface. The node density along the wave propagation direction is set lower than that of the SAW structure because of the lower resonance frequency of QCM (∼10 MHz) compared to that of SAW (∼132 MHz). To calculate the frequency shifts of the sensors due to mercury exposure, both SAW and QCM structures were simulated for unperturbed and also mercury-perturbed Au electrodes, and their resonance frequencies were compared. In the first set of simulations, it was assumed that no Hg was present on the Au electrodes and the unperturbed Au density (19.3 g/cm3) was used while the thickness of the Au electrodes was kept at 100 nm. Figure 2a,b illustrates the surface plot of the total displacement (defined as pzd.dsp in COMSOL multiphysics) at resonance frequencies for SAW- and QCMbased sensors under the electrode area. A second set of simulations were performed by considering the changes in Au electrode properties as a result of Hg exposure. It was assumed that the sensors were exposed to a Hg0 vapor concentration of 113 ppbv which is calculated to deposit the equivalent of 0.025 of a Hg monolayer on the Au surface.27 The effective mass density and the thickness of Hg-perturbed Au electrodes were calculated by assuming each Hg monolayer has a surface coverage of 469 ng/cm2 and the Hg atom diameter is 0.342 nm.28,29 The effective mass density and the thickness of the Au C
DOI: 10.1021/acs.langmuir.5b01858 Langmuir XXXX, XXX, XXX−XXX
Article
Langmuir
Figure 3. Dynamic response of (a) SAW- and (b) QCM-based sensors toward different concentrations of Hg0 vapor at 35 and 55 °C. Comparison of normalized responses of SAW- and QCM-based sensors for different concentrations of Hg0 vapor at (c) 35 and (d) 55 °C. and insertion loss of the sensor were ∼131 MHz and −11 dB respectively as measured using a vector network analyzer (Rhode & Schwartz). An AT cut quartz crystal of 7.5 mm diameter and 166 μm thickness was utilized to fabricate the QCM device. A 10 nm Ti adhesion layer followed by a 100 nm Au thin film was deposited on both sides of the substrate using an evaporation method similar to that used for the SAW device. The circularly shaped Au electrodes on both sides of the device were patterned by employing a shadow mask. The total mechanical surface areas of Au on the SAW and QCM devices were 0.073 and 0.32 mm2, respectively. 3.2. Experimental Setup for Hg0 Vapor Testing. The sensors were kept inside a chamber (∼100 mL volume) under a dry nitrogen (N2) atmosphere in order to stabilize them prior to any Hg0 vapor exposure. Different concentrations of Hg0 vapor ranging from 24 to 365 ppbv were generated by controlling the temperature (40 to 80 °C) of a NIST-certified permeation tube (VICI, TX, USA) with a target rate of 3100 ng/min at 100 °C (as certified by NIST). To reconfirm the concentrations of the generated Hg0 vapor, the gas stream was first trapped in a series of four potassium permanganate (KMnO4) solutions just before the gas entered the sensor chamber and then analyzed using inductively coupled plasma mass spectrometry (ICPMS). The sensor chamber temperature was controlled using a heater that utilized thermocouples inside the chamber as feedback. A multichannel mass flow controller (MFC) system was utilized to control the flow of Hg0 vapor, H2O vapor, and the different interfering gases tested. The interfering gas species tested were ammonia (NH3), acetaldehyde (MeCHO), ethylmercaptan (EM), dimethyl disulfide (DMDS), and methyl ethyl ketone (MEK), each supplied in separate cylinders. H2O vapor was generated using a V-Gen humidity generator (InstruQuest Inc.). A total flow of 200 standard cubic centimeters (sccm) per minute was kept constant during the entire experiment. In this study, the Hg0 vapor concentrations utilized were 24, 51, 104, 195, 265, and 365 ppbv. The concentrations of NH3, MeCHO, EM, DMDS, and MEK tested were 383.8, 303.4, 2.61, 5.01, and 40.1 ppmv, respectively, while H2O vapor was generated at 20% relative humidity
electrode for Hg0 exposure were calculated to be 19.30115 g/ cm3 and 0.1000086 μm, respectively. The surface plots of the total displacement at resonance frequencies under the electrode area of SAW- and QCM-based sensors perturbed by Hg can be seen in Figure 2c,d. As shown in Figure 2e, the resonance frequency of the SAW-based sensor shifted from 131 977 210 to 131 977 150 Hz (60 Hz reduction) when the Au electrodes were affected by the Hg0 content of 113 ppbv (∼0.025 monolayer). It can be seen from Figure 2f that the QCM-based sensor showed approximately 0.3 Hz (resonance frequency shifted from 10 054 596.2 to 10 054 595.9 Hz) for the same Hg content in spite of having a 4 times larger surface area, thereby showing that the sensitivity of the SAW is ∼200 times higher than that of the QCM-based sensor for the same concentration of Hg0 vapor.
3. EXPERIMENTAL SECTION 3.1. Device Fabrication. The SAW device used in this study was fabricated by patterning Au electrodes on a ST-cut quartz substrate. ST-cut quartz is a commonly used substrate for fabricating SAW-based devices because of its relatively higher temperature stability at 20−70 °C when compared to materials such as XY-lithium niobate or YXlithium tantalate.30 First, 20 and 30 nm titanium (Ti) and nickel (Ni) adhesion layers were deposited on top of an ST-quartz substrate (15 mm in length, 9 mm in width and 500 μm in thickness), respectively. Then a 50 nm Au thin film was deposited on top of the adhesion layers. Metal depositions were performed using a Balzers e-beam (BAK 600) evaporator at 22 °C. The Au electrodes (input and output IDTs) were then patterned by using a combination of standard photolithography and wet-etching processes. A total of 360 electrode pairs (180 in each IDT with a 75 μm edge-to-edge separation between input and output IDTs) were patterned where each electrode had a width of 6 μm (acoustic wavelength of 24 μm). A 1700 μm aperture width (w) was maintained for all electrodes. The resonance frequency D
DOI: 10.1021/acs.langmuir.5b01858 Langmuir XXXX, XXX, XXX−XXX
Article
Langmuir Table 1. Response Magnitudesa
(55 °C) and utilized for selectivity testing of the developed sensors. The performance of the sensors was tested at different temperatures, ranging from 35 to 75 °C with an increment of 20 °C in order to investigate the sensors’ sorption and desorption behavior at different operating temperatures. The sensor chamber was placed in a fume cupboard where the internal temperature varies between 28 and 34 °C because of the air flow. Therefore, 35 °C was set as the lowest operating temperature in order to prevent the sensors’ response profile from being affected by the temperature fluctuations in the fume cupboard. The operating temperatures of 55 and 75 °C were chosen as they resemble the operating temperatures of most industrial processes (i.e., alumina refineries). Following a 30 min exposure to Hg0 vapor, the sensors were exposed to dry nitrogen (N2) for a period of 90 min in order to desorb the mercury molecules from the Au surface and regenerate the sensor. This total period (30 min of Hg exposure and 90 min of N2 purging) is referred to as a pulse throughout the article. An rf amplifier was fabricated and used to create the SAW oscillatory circuit, and an Agilent frequency counter was used to monitor the oscillation frequency of the SAW-based sensor. A Maxtek RQCM (10 MHz phase-locked oscillator, Beaverton, OR) was used to track the oscillation frequency of the QCM-based sensor. The RQCM had a resoulution of ±0.01 Hz. The measuring apparatus (i.e., RQCM) was calibrated before running the experiments in order to nullify the effects that they have on the sensors’ response profiles. 3.3. Hg0 Depth Profile. X-ray photoelectron spectroscopy (XPS) of the surface and depth profile analyses were performed using a Thermo Scientific K-Alpha instrument with a monochromated Al Kα radiation source of 1486.7. A low-energy Ar plasma flood gun was used to help alleviate surface charging, and the core levels were charge shifted to align the C 1s binding energy to 284.8 eV from adventitious carbon. The spectra components were background corrected using the Shirley algorithm and deconvoluted using a Gaussian−Lorentzian function considering least-squares fitting. Depth profiling was undertaken using successive argon ion etching processes while monitoring the spectroscopy of core levels for Au 4f, Hg 4f, and Ni 2p binding energies. Atomic force microscopy (AFM) characterization was performed in order to determine the depth of the crater that was formed during the argon ion etching process. AFM images and height profile along the IDTs of the SAW device were obtained using a Digital Instruments D3100 atomic force microscope.
Hg0concentration (ppbv)
35 °C
55 °C
75 °C
24 51 104 195 265 365
110.7 94.4 94.6 89.3 87.2 87.8
35.2 33.2 38.6 35.9 35.0 35.0
20.1 15.5 17.4 18.5 18.9 20.0
a Δf SAW (Hz): Δf QCM (Hz) ratio of SAW- and QCM-based sensors toward different concentrations of Hg0 vapor at various operating temperatures.
sensor rather than in the bulk, which is the case for a QCMbased sensor. The higher resonance frequency (f 0) is also responsible for the higher sensitivity of the SAW-based sensor as the change in resonance frequency (Δf) of both SAW- and QCM-based sensors (due to Hg0 vapor exposure) is directly proportional to the resonance frequency of the respective sensor (eqs 1 and 2).21 The difference between the SAW- and QCM-based sensor response magnitudes was found to decrease with the increase in operating temperature. That is, the SAWbased sensor exhibited ∼33 to ∼39 and ∼16 to ∼20 times higher response magnitudes toward different concentrations of Hg0 vapor at operating temperatures of 55 and 75 °C, respectively. The stability of the baseline (no Hg exposure condition) frequency of SAW- and QCM-based sensors at 35 and 55 °C can also be observed in Figure 3a,b, respectively. It can be seen in Figure 3a that the drift from the baseline frequency (the residual frequency after desorption in N2) of the SAW-based sensor was reduced when the operating temperature was increased from 35 to 55 °C; however, no significant improvement in the baseline frequency drift was observed for the QCM-based sensor when the operating temperature was changed from 35 to 55 °C. Figure 3c,d shows the comparison of the normalized response curves for both the SAW- and QCM-based sensors at 35 and 55 °C, respectively. It can be observed that the baseline frequency of the SAW-based sensor was more stable than that of QCM-based sensor for both operating temperatures, indicating that the SAW-based sensor had a better desorption efficiency. The desorption efficiency of the sensors can be estimated by calculating the ratio (in percent) of desorption to adsorption response magnitudes toward Hg0 vapor. The SAW-based sensor showed a 94 to 100% desorption efficiency for a Hg0 vapor concentration of 265 ppbv at the operating temperatures tested, and the QCMbased sensor exhibited a desorption rate of 78 to 95% under the same conditions. 4.2. Hg0 Sorption and Desorption Isotherms. Figure 4a,b illustrates the sorption and desorption calibration curves, respectively, (response magnitudes vs Hg0 vapor concentrations) of both SAW- and QCM-based sensors at an operating temperature of 35 °C, and Figure 4c,d shows the sorption and desorption calibration curves of the sensors when operating at 55 °C. It can be observed from Figure 4a,c that the sorption response magnitudes of both sensors toward different concentrations of Hg0 vapor for sorption followed the Langmuir extension isotherm at both operating temperatures of 35 and 55 °C. The coefficient of determination (R2) of the Langmuir fit for the SAW- and QCM-based sensor was found to be 0.9968 and 0.9940 for SAW- and QCM-based sensors, respectively, at an operating temperature of 35 °C, and R2
4. RESULTS AND DISCUSSION 4.1. Sensor Performance. A series of experiments were undertaken to analyze the performance of both the SAW- and QCM-based sensors toward Hg0 vapor detection at a range of operating temperatures (as stated in section 3.2). A negative shift in the resonance frequency ( f 0) was observed for both of the sensors upon exposure to Hg0 vapor, and they were found to return to their baseline frequency as Hg0 desorption occurred. Figure 3a,b shows the developed SAW- and QCMbased sensors’ response toward different concentrations of Hg0 vapor, respectively, at operating temperatures of 35 and 55 °C. It can be seen that the response magnitudes for each sensor increased with increasing Hg0 vapor concentration. It can also be observed from Figure 3a,b that the response magnitudes of the SAW-based sensor toward Hg0 vapor decreased when the operating temperature increased from 35 to 55 °C while the opposite phenomenon was observed in the case of the QCMbased sensor. However, the SAW-based sensor showed a relatively higher sensitivity toward Hg0 vapor over the QCMbased sensor at all operating temperatures. As stated in Table 1, the SAW-based sensor showed ∼88 to ∼111 times higher response magnitudes than the QCM-based sensor at various concentrations of Hg0 vapor while operating at 35 °C. This significantly higher response magnitude of the SAW-based sensor compared to that of the QCM counterpart was observed as more energy is confined on the surface of a SAW-based E
DOI: 10.1021/acs.langmuir.5b01858 Langmuir XXXX, XXX, XXX−XXX
Article
Langmuir
Figure 4. Langmuir extension isotherm fit of the calibration curves for SAW- and QCM-based sensors for (a) Hg0 sorption and (b) desorption data at an operating temperature of 35 °C and (c) Hg0 sorption and (d) desorption data at an operating temperature of 55 °C.
QCM-based sensors are slightly lower than the sorption magnitudes. This can be attributed to the diffused/amalgamated Hg0 molecules that require high-energy input in order to be removed from the Au electrodes.31 However, the high desorption to sorption ratios calculated (94 to 100% for the SAW-based sensor and 78 to 95% for the QCM-based sensor) for such high external energy input (i.e., heat treatment) are not necessary for these mass-based sensors. 4.3. Hg0 Sorption and Desorption Rate. The sorption and desorption rates of the sensors were calculated by differentiating the sensors’ dynamic responses with respect to time for each concentration of Hg0 vapor tested.31 Figure 5a,b illustrates the sorption rate of the SAW- and QCM-based sensors, respectively, for a range of concentrations of Hg0 vapor while operating at 35 °C. As expected, the SAW-based sensor showed a much higher sorption rate (up to ∼68 times) than the QCM-based sensor for each concentration of Hg0 vapor exposed. It was also observed that the SAW- and QCM-based sensors reached the maximum sorption rate at only ∼67 and ∼72 s, respectively when exposed to a Hg0 vapor of 24 ppbv. However, the time needed to reach the maximum sorption rate decreased more rapidly when the exposed Hg0 vapor concentrations were increased. As observed, it took only ∼21 s for the SAW-based sensor to reach the maximum sorption rate for 365 ppbv exposure Hg0 at 35 °C. Similarly, only ∼42 s was required for the QCM-based sensor to reach the maximum sorption rate under similar conditions. It can also be observed from Figure 5a,b that the drop in the sorption rate from the maximum point to zero became slower as the concentrations of the Hg0 vapor increased in the case of both sensors, indicating
values of 0.9817 (SAW) and 0.9955 (QCM) were observed when the sensors operated at 55 °C. The relation between the response magnitudes (resonance frequency shift, Δf) and the exposed Hg0 vapor concentration ([Hg]) for both sensors can be determined from the Langmuir extension isotherm (eq 3), which can be further differentiated to determine the sensitivity of the sensors16 (eq 4). ab[Hg]c (1 + b[Hg]c )
(3)
abc[Hg]c − 1 df = d[Hg] (1 + b[Hg]c )2
(4)
Δf =
In eqs 3 and 4, a, b, and c are constants that depend on the sensor type and operating temperatures. It can be observed from eq 4 that the sensitivity of both sensors is dependent on the Hg0 vapor concentration, with a decrease in sensitivity when the Hg0 vapor concentration is increased, which indicates that both the developed SAW- and QCM-based sensors are well suited for detecting lower concentrations of Hg0 vapor. It can be observed from Figure 4b that the positive frequency shift which results from Hg0 desorption from the Au surface at 35 °C are also well fitted with the Langmuir extension isotherm for the SAW-based sensor (R2 value of 0.9950); however, the QCM-based sensor exhibited a poor fit with an R2 value of 0.7855. On the other hand, the frequency shifts of both sensors for Hg0 desorption at 55 °C were found to be a good fit, with R2 values of 0.9981 and 0.9983 for SAW- and QCM-based sensor, respectively (Figure 4d). One can also observe from Figure 4a−d that the desorption magnitudes of the SAW- and F
DOI: 10.1021/acs.langmuir.5b01858 Langmuir XXXX, XXX, XXX−XXX
Article
Langmuir
Figure 5. Hg0 sorption rate for different concentrations of Hg0 vapor demonstrated by (a) SAW- and (b) QCM-based sensors at 35 °C. (c) Langmuir extension isotherm fit for calibration curves using the maximum sorption rate data demonstrated by both sensors at different Hg0 vapor concentrations and sorption rate demonstrated by the (d) SAW- and (e) QCM-based sensors following 265 ppbv of Hg0 vapor exposure at three different operating temperatures,.
that Hg0 sorption/amalgamation on the Au surface increases with the concentration of Hg0 vapor. The diffusion of the Hg0 atoms into the bulk of the Au surface is also expected to increase with the increase in Hg0 vapor concentrations as diffusion is a concentration-gradient-dependent process (further discussion in section 4.5). It can be observed from Figure 5c that the maximum adsorption rate demonstrated by the sensors for different concentrations of Hg0 vapor fits well with the Langmuir extension isotherm. Therefore, eq 3 can be used to correlate the maximum sorption rate with the Hg0 vapor concentration, thereby reducing the sensor response time to seconds rather than minutes. Sorption rates were also found to be dependent on the operating temperature for both the SAW- and the QCM-based sensors. Figure 5d,e shows the Hg0 sorption rate of SAW- and QCM-based sensors, respectively, at three operating temperatures and a Hg0 vapor concentration of 265 ppbv. It can be observed that the Hg0 sorption rate for the SAW-based sensor decreased with increasing operating temperature, which was opposite to the trend observed for the QCM-based Hg0 vapor sensor. The times required to reach the maximum sorption rate for the SAW-based sensor were observed to be ∼25, ∼ 32, and ∼33s at 35, 55, and 75 °C, respectively, indicating that the operating temperature has little effect on this parameter. However, in the case of the QCM-based sensor the time required to reach the maximum sorption rate was found to increase with increasing operating temperature, with maximum times of ∼48, ∼55, and ∼78 s being required at 35, 55, and 75 °C, respectively. It can also be observed from Figure 5c,d that the time required by the SAW-based sensor’s sorption rate to
drop from the maximum value to zero increased with increasing operating temperature whereas the opposite phenomenon was noticed for the QCM-based sensor. The desorption rates of SAW- and QCM-based sensors following the exposure of a range of Hg0 vapor concentrations at 35 °C can be seen in Figure 6a,b, respectively. These maximum desorption rates were found to correlate with the exposed Hg0 vapor concentrations and fit well with the Langmuir extension isotherm as shown in Figure 6c. As expected, the desorption rate for both sensors increased with the tested Hg0 vapor concentrations. It was observed that the SAW-based sensor required only 42 to 78 s to reach the maximum desorption rate, where the time decreased with increasing Hg0 vapor concentrations. The QCM-based sensor also exhibited a similar trend and required 48−67 s to reach the maximum desorption rate for different concentrations of Hg0 vapor. The temperature dependency of the Hg0 desorption rate for both SAW- and QCM-based sensors can be observed in Figure 6d,e, respectively. The maximum desorption rate for the SAW-based sensor was found to decrease with increasing operating temperature. However, an increase in operating temperature increased the Hg0 desorption rate for the QCMbased sensor. The opposite trend of SAW- and QCM-based sensors’ sorption rates with temperature change is attributed to the different piezoelectric substrate cut used to fabricate each sensor. That is, ST-cut quartz was used to fabricate SAW while the QCM-based sensor was fabricated using AT-cut quartz as was required. Each quartz cut is well known to possess different temperature coefficients.32 G
DOI: 10.1021/acs.langmuir.5b01858 Langmuir XXXX, XXX, XXX−XXX
Article
Langmuir
Figure 6. Hg0 desorption rate for different concentrations of Hg0 vapor demonstrated by (a) SAW and (b) QCM-based sensors at 35 °C. (c) Langmuir extension isotherm fit for calibration curves using the maximum desorption rate data demonstrated by both sensors at different Hg0 vapor concentrations and the desorption rate demonstrated by the (d) SAW- and (e) QCM-based sensors following 265 ppbv of Hg0 vapor exposure at three different operating temperatures.
Figure 7. Response time (t90) of the SAW- and QCM-based sensors for 265 ppbv Hg0 vapor exposure at operating temperatures of (a) 35 and (b) 55 °C. (c) Comparison of t90 obtained from the response magnitude and sorption rate data of the sensors.
4.4. Response Time. To determine the response time for both the developed SAW- and QCM-based sensors at different operating temperatures, the t90 value was calculated for 265 ppbv Hg0 exposure at 35 and 55 °C. The term t90 is defined as the time required to reach 90% of the response signal of the equilibrium value of a certain event. These values were calculated from the dynamic response magnitudes obtained directly from the sensors’ data as well as for the calculated sorption rates. Figure 7a,b shows the t90 of the developed sensors calculated from the sensors’ dynamic response toward a Hg0 vapor pulse of 265 ppbv. It can be observed that both sensors had similar t90 values of ∼1440 and ∼1380 s at 35 and 55 °C, respectively. The results indicated that the response time
The ratio of maximum desorption rate to maximum sorption rate increased for the SAW-based sensor as the operating temperature increased (0.38 and 0.49s at 35 and 55 °C, respectively, for 295 ppbv of Hg0 vapor); however, this ratio was found to decrease with the operating temperature for the developed QCM-based sensor (0.39 and 0.37 s at 35 and 55 °C, respectively). Overall, it can be observed that the maximum desorption rate of a certain Hg0 vapor pulse is significantly lower than the corresponding sorption rate for both SAW- and QCM-based sensors regardless of the operating temperature. This indicates that the strong affinity and amalgamation process between Hg0 atoms and the Au surface makes the desorption process slower than the sorption process. H
DOI: 10.1021/acs.langmuir.5b01858 Langmuir XXXX, XXX, XXX−XXX
Article
Langmuir
Figure 8. XPS spectra of (a) Au 4f and (b) Hg 4f core levels for Hg-exposed Au electrodes deposited on the SAW-based sensor. (c) XPS depth profiles of Au, Hg, and Ni of the Au electrode analyzed for atomic percentage and (d) AFM image of a portion of the IDT showing four of the electrodes of SAW devices as well as the thickness profile of Au electrodes prior to Hg0 vapor exposure. The AFM data revealed that for the XPS depth profile shown in (c), 57 etch level = 50 nm indicates that mercury had diffused ∼30 nm into the Au surface of the sensor.
the nondiffused Hg0 molecules adsorbed on the Au electrode surface, which were removed from the surface because of the low operating pressure of the XPS.36 Then the mercury concentration decreases at each incremental stage of the depth profiles in the sample because of the diffusion of the mercury into the IDT. It can be observed that Hg diffused through approximately half of the Au electrode thickness (white area). In the last section (magenta shaded area) of the depth profile, the Au atomic percentage started to decrease drastically as Ni core-level peaks appeared in the depth profile, indicating the complete etching of the Au layer of the electrode. AFM characterization was performed in order to determine the depth at which the argon ion beam etched through the Au surface at each stage. It was found that the total 57 etch levels referred to the Au layer of 50 nm thickness, which indicates that the mercury diffused a total of ∼30 nm through the Au surface. This is much deeper than the previously reported study in the literature37 indicating that Hg can diffuse to great lengths into Au films with long exposure to Hg0 vapor.38 The AFM image and the corresponding height profile show the formation of even electrodes with 100 nm thickness on the SAW device as shown in Figure 8d, corresponding to the Au layer and underlying Ni and Ti adhesion layers on the substrate. 4.6. Effect of Interferants on the Sensors’ Response. To test the selectivity of the developed SAW- and QCM-based sensors toward a low concentration of Hg0 vapor while in the
of the developed sensors did not vary significantly with the operating temperatures tested. On the other hand, the t90 for both sensors was found to decrease significantly when obtained from the sorption rate data, as can be seen from Figure 7c. The SAW-based sensor had t90 values of 8 and 17 s for 265 ppbv Hg0 vapor at 35 and 55 °C, respectively, and the QCM-based sensor had t90 values of 34 and 38 s for the same conditions. These values indicate that relative to the sensor response data, the sorption rate data can be utilized to obtain 180 and 40 times faster response times for the SAW- and QCM-based sensors, respectively, at 35 °C. On the other hand, 85 and 37 times faster response times can be obtained at an increased operating temperature of 55 °C. 4.5. Hg0 Depth Profile. To determine the extent of diffused Hg into the Au surface, XPS depth profiling was employed. Figure 8a,b shows the core-level binding energies of Au 4f and Hg 4f of the Au IDT electrode on the SAW-based sensor after Hg0 vapor experiments. The Au 4f7/2 binding energy at 84.2 eV is related to Au0 core levels, and the Hg 4f7/2 core binding energy at 100.4 eV is attributed to metallic mercury.9,15,33−35 The depth profile of the IDT in the SAWbased sensor is shown in Figure 8c. The first (blue shaded area) section of the depth profile shows the atomic percentage of the different metal species on the surface of the Au electrode. It is interesting to observe that the atomic percentage of Hg rose to a peak before decreasing to a stable level. This may be due to I
DOI: 10.1021/acs.langmuir.5b01858 Langmuir XXXX, XXX, XXX−XXX
Article
Langmuir
Figure 9. Dynamic response of the (a) SAW and (b) QCM-based sensors exposed to 265 ppbv Hg0 vapor with/without the presence of different interfering gas species at an operating temperature of 55 °C. (c) Bar graph showing the overall selectivity of the sensors’ toward Hg0 vapor in the presence of various interfering gases while operating at 55 °C.
a low concentration (low ppbv levels) of Hg0 vapor. Detailed experiments and mathematical analysis were performed in order to determine the sensitivity, sorption rate, desorption rate, and response time of both sensors. The SAW-based sensor was found to have a higher sensitivity toward Hg0 vapor than the QCM-based sensor, showing up to ∼111 and ∼39 times higher response magnitudes at 35 and 55 °C, respectively. A decreasing trend in the maximum sorption and desorption rate was observed for the SAW-based sensor, but the QCM-based sensor demonstrated the opposite trend. In general, the SAWbased sensor was found to reach the maximum sorption rate faster; however, the response time (t90 obtained from the dynamic response profiles) toward Hg0 vapor was found to be similar for both sensors. The results also indicated that the sorption and desorption rate can be utilized as an alternative to the dynamic response magnitude, which can potentially lead to concentrations of Hg0 vapor being detected 180 times faster. The XPS depth profile of the Au surface following Hg0 vapor exposure experiments revealed that mercury had diffused beyond the 6 nm (reported in the literature) as a result of a longer exposure period associated with this study. Finally, the SAW-based sensor was found to have slightly higher overall selectivity (90%) than the QCM counterpart (84%), and Hg0 vapor was measured in the presence of ammonia (NH3), humidity, and volatile organic compounds.
presence of common interfering gas species, repeated pulses of Hg0 vapor were exposed to the sensors along with an interfering gas added to the alternative pulses. Figure 9a,b corresponds to the SAW- and QCM-based sensors’ dynamic response, respectively, when exposed to 265 ppbv of Hg0 vapor with and without the presence of NH3, MeCHO, EM, DMDS, MEK, and H2O vapor at an operating temperature of 55 °C. It can be observed that the response magnitudes of each sensor did not deviate significantly when Hg0 vapor was exposed to the sensors with/without the presence of each interfering gas species. This can be better observed in Figure 9c, where it is demonstrated that the SAW-based sensor lost only up to 10% of its response magnitude (i.e., 90% selectivity23) when an interfering gas was present compared to Hg0 vapor exposure alone. However, the QCM-based sensor was found to lose up to 16% of its response magnitude (i.e., 84% selectivity) for the same conditions, indicating that the SAW-based sensor possess a relatively higher selectivity toward low concentrations of Hg0 vapor than the QCM counterpart. The high affinity of Hg0 molecules toward the Au electrode surface, which forms a strong Hg−Au amalgam, is attributed to the high selectivity of the sensors toward low concentrations of Hg0 vapor.
5. CONCLUSIONS In this study, a comparative analysis of SAW- and QCM-based Hg0 vapor sensing was presented. First, an FEM model was presented where it was observed that a SAW-based sensor with 25% surface area of a QCM-based sensor’s sensitive layer can exhibit an up to ∼200 times higher response magnitude toward
■
AUTHOR INFORMATION
Corresponding Authors
*E-mail:
[email protected]. J
DOI: 10.1021/acs.langmuir.5b01858 Langmuir XXXX, XXX, XXX−XXX
Article
Langmuir *E-mail:
[email protected].
(17) Jakubik, W. P. Surface acoustic wave-based gas sensors. Thin Solid Films 2011, 520, 986. (18) Hsu, C.-L.; Shen, C.-Y.; Tsai, R.-T.; Su, M.-Y. Surface acoustic wave ammonia sensors based on ST-cut quartz under periodic Al structure. Sensors 2009, 9 (2), 980−994. (19) Ballantine, D. S., Jr; Wohltjen, H. Surface acoustic wave devices for chemical analysis. Anal. Chem. 1989, 61 (11), 704A−715A. (20) Wohltjen, H. Mechanism of operation and design considerations for surface acoustic wave device vapour sensors. Sens. Actuators 1984, 5 (4), 307−325. (21) Ballantine Jr, D.; White, R. M.; Martin, S. J.; Ricco, A. J.; Zellers, E.; Frye, G.; Wohltjen, H. Acoustic Wave Sensors: Theory, Design, & Physico-Chemical Applications; Academic Press, 1996. (22) Kanazawa, K.; Cho, N.-J. Quartz crystal microbalance as a sensor to characterize macromolecular assembly dynamics. J. Sens. 2009, 2009.110.1155/2009/824947 (23) Kabir, K. M.; Sabri, Y. M.; Matthews, G. I.; Jones, L. A.; Ippolito, S.; Bhargava, S. K. Selective detection of elemental mercury vapor using a surface acoustic wave (SAW) sensor. Analyst 2015, DOI: 10.1039/C5AN00360A. (24) Sauerbrey, G. Use of quartz vibration for weighing thin films on a microbalance. Eur. Phys. J. A 1959, 155, 206−212. (25) Hoummady, M.; Campitelli, A.; Wlodarski, W. Acoustic wave sensors: design, sensing mechanisms and applications. Smart Mater. Struct. 1997, 6 (6), 647. (26) Auld, B. A. Acoustic Fields and Waves in Solids; John Wiley, 1973. (27) Sabri, Y. M.; Ippolito, S. J.; Tardio, J.; Atanacio, A. J.; Sood, D. K.; Bhargava, S. K. Mercury diffusion in gold and silver thin film electrodes on quartz crystal microbalance sensors. Sens. Actuators, B 2009, 137 (1), 246−252. (28) Morris, T.; Sun, J.; Szulczewski, G. Measurement of the chemical and morphological changes that occur on gold surfaces following thermal desorption and acid dissolution of adsorbed mercury. Anal. Chim. Acta 2003, 496 (1−2), 279−287. (29) Haskell, R. B. A Surface Acoustic Wave Mercury Vapor Sensor. The University of Maine, 2003. (30) Yamazaki, T.; Iizawa, K.; Kanna, S.; Takagi, M. Temperature stability of surface acoustic wave resonators on in-plane rotated 33 Ycut Quartz. Jpn. J. Appl. Phys. 2003, 42 (5S), 3136. (31) Sabri, Y. M.; Ippolito, S. J.; Tardio, J.; Bhargava, S. K. Study of Surface Morphology Effects on Hg Sorption−Desorption Kinetics on Gold Thin-Films. J. Phys. Chem. C 2012, 116 (3), 2483−2492. (32) Caron, J. J.; Andle, J.; Vetelino, J. F. Surface Acoustic Wave Substrates for High Temperature Applications; IEEE International Frequency Control Symposium, Honolulu, HI, 1996; IEEE, pp 222−227. (33) Esmaielzadeh Kandjani, A.; Sabri, Y. M.; Mohammad-Taheri, M.; Bansal, V.; Bhargava, S. K. Detect, Remove and Reuse: A New Paradigm in Sensing and Removal of Hg (II) from Wastewater via SERS-Active ZnO/Ag Nanoarrays. Environ. Sci. Technol. 2015, 49, 1578. (34) Ren, W.; Zhu, C.; Wang, E. Enhanced sensitivity of a direct SERS technique for Hg2+ detection based on the investigation of the interaction between silver nanoparticles and mercury ions. Nanoscale 2012, 4 (19), 5902−5909. (35) George, M. A.; Glaunsinger, W. S.; Thundat, T.; Lindsay, S. M. Investigation of mercury adsorption on gold films by STM. J. Microsc. 1988, 152 (3), 703−713. (36) Hutson, N. D.; Attwood, B. C.; Scheckel, K. G. XAS and XPS characterization of mercury binding on brominated activated carbon. Environ. Sci. Technol. 2007, 41 (5), 1747−1752. (37) Battistoni, C.; Bemporad, E.; Galdikas, A.; Kačiulis, S.; Mattogno, G.; Mickevičius, S.; Olevano, V. Interaction of mercury vapour with thin films of gold. Appl. Surf. Sci. 1996, 103 (2), 107−111. (38) Sabri, Y. M.; Ippolito, S. J.; Al Kobaisi, M.; Griffin, M. J.; Nelson, D. R.; Bhargava, S. K. Investigation of Hg sorption and diffusion behavior on ultra-thin films of gold using QCM response analysis and SIMS depth profiling. J. Mater. Chem. 2012, 22 (39), 20929−20935.
Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS We acknowledge the Microelectronic and Materials Technology Centre (MMTC) at RMIT University for allowing the use of their facilities. We also acknowledge the Australian Research Council (ARC) for supporting this project, and S.J.I. acknowledges the ARC for an APDI fellowship (LP100200859). We thank the RMMF (RMIT Microscopy and Microanalysis Facility) for their equipment and expertise. We also thank King Saud University for supporting this work via research group no. RGP-VPP-236.
■
REFERENCES
(1) Qiu, J. Tough talk over mercury treaty. Nature 2013, 493, 144− 145. (2) Syversen, T.; Kaur, P. The toxicology of mercury and its compounds. J. Trace Elem. Med. Biol. 2012, 26 (4), 215−226. (3) Holmes, P.; James, K. A. F.; Levy, L. S. Is low-level environmental mercury exposure of concern to human health? Sci. Total Environ. 2009, 408 (2), 171−182. (4) Burger Chakraborty, L.; Qureshi, A.; Vadenbo, C.; Hellweg, S. Anthropogenic mercury flows in India and impacts of emission controls. Environ. Sci. Technol. 2013, 47 (15), 8105−8113. (5) Liang, S.; Zhang, C.; Wang, Y.; Xu, M.; Liu, W. Virtual Atmospheric Mercury Emission Network in China. Environ. Sci. Technol. 2014, 48, 2807. (6) Hatch, W. R.; Ott, W. L. Determination of submicrogram quantities of mercury by atomic absorption spectrophotometry. Anal. Chem. 1968, 40 (14), 2085−2087. (7) Granite, E. Obstacles in the Development of Mercury Continuous Emissions Monitors. Proceedings of 20th Annual International Pittsburgh Coal Conference, 2003; pp 30−31. (8) Selid, P. D.; Xu, H.; Collins, E. M.; Striped Face-Collins, M.; Zhao, J. X. Sensing mercury for biomedical and environmental monitoring. Sensors 2009, 9 (7), 5446−5459. (9) Sabri, Y. M.; Ippolito, S. J.; Tardio, J.; Bansal, V.; O’Mullane, A. P.; Bhargava, S. K. Gold nanospikes based microsensor as a highly accurate mercury emission monitoring system. Sci. Rep. 2014, 4, 6741. (10) Laudal, D. L.; Thompson, J. S.; Pavlish, J. H.; Brickett, L. A.; Chu, P. Use of continuous mercury monitors at coal-fired utilities. Fuel Process. Technol. 2004, 85 (6−7), 501−511. (11) Fitzgerald, W. F.; Gill, G. A. Subnanogram determination of mercury by two-stage gold amalgamation and gas phase detection applied to atmospheric analysis. Anal. Chem. 1979, 51 (11), 1714− 1720. (12) James, J. Z.; Lucas, D.; Koshland, C. P. Gold nanoparticle films as sensitive and reusable elemental mercury sensors. Environ. Sci. Technol. 2012, 46 (17), 9557−9562. (13) McNicholas, T. P.; Zhao, K.; Yang, C.; Hernandez, S. C.; Mulchandani, A.; Myung, N. V.; Deshusses, M. A. Sensitive detection of elemental mercury vapor by gold-nanoparticle-decorated carbon nanotube sensors. J. Phys. Chem. C 2011, 115 (28), 13927−13931. (14) Drelich, J.; White, C. L.; Xu, Z. Laboratory tests on mercury emission monitoring with resonating gold-coated silicon cantilevers. Environ. Sci. Technol. 2008, 42 (6), 2072−2078. (15) Sabri, Y. M.; Esmaielzadeh Kandjani, A.; Ippolito, S. J.; Bhargava, S. K. Nanosphere Monolayer on a Transducer for Enhanced Detection of Gaseous Heavy Metal. ACS Appl. Mater. Interfaces 2015, 7 (3), 1491−1499. (16) Kabir, K. M.; Sabri, Y. M.; Matthews, G. I.; Ippolito, S. J.; Bhargava, S. K. Cross sensitivity effects of volatile organic compounds on a SAW-based elemental mercury vapor sensor. Sens. Actuators, B 2015, 212, 235−241. K
DOI: 10.1021/acs.langmuir.5b01858 Langmuir XXXX, XXX, XXX−XXX
