Anal. Chem. 2007, 79, 251-255
Piezoelectric Quartz Crystal Microbalance Sensor for Trace Aqueous Cyanide Ion Determination Yegor G. Timofeyenko, Jeffrey J. Rosentreter,* and Susan Mayo
Downloaded via OPEN UNIV OF HONG KONG on January 23, 2019 at 18:23:54 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
Department of Chemistry, Idaho State University, Pocatello, Idaho 83209
Using selective reaction chemistry, our present research has developed an online, real-time sensor capable of monitoring toxic cyanide at both drinking water standard and environmental regulatory concentrations. Through the use of a flow cell, aqueous samples containing cyanide are reacted with a gold electrode of a piezoelectric crystal to indirectly sense cyanide concentration by the dissolution of metallic gold. The quartz crystal is an AT-cut wafer sandwiched between two neoprene O-rings within the liquid flow cell. The presence of cyanide in solution results in the selective formation of a soluble dicyano-gold complex according to the Elsner reaction: 4Au + 8CN+ 2H2O + O2 a 4Au(CN)2- + 4OH-. The resulting loss of gold from the electrode is detected by the piezoelectric crystal as a resonant frequency change. Since free cyanide is a weak acid (pKa ) 9.3), available protons compete for cyanide ligands. Therefore, increased sample pH provides higher sensitivity. The detection limits at pH 12 are 16.1 and 2.7 ppb for analysis times of 10 min and 1 h, respectively. The incorporation of the flow cell improves both analyte sensitivity and instrument precision, with an average signal intensity drift of only 5% over a 2-h analysis. The calibrations show excellent linearity over a variety of cyanide concentrations ranging from low ppb to hundreds of ppm. This detection method offers the advantage of selectively detecting cyanides posing a biohazard while avoiding detection of stable metal cyanides. This aspect of the system is based on competitive exchange of available metals and gold with cyanide ligands. Stable metal cyanide complexes possess a higher formation constant than cyanoaurate. This detection system has been configured into a flow injection analysis array for simple adaptation to automation. Anions commonly found in natural waters have been examined for interference effects. Additionally, the sensor is free from interference by aqueous cyanide analogues including thiocyanate. The developed detection system provides rapid cyanide determinations with little sample preparation or instrument supervision. Cyanide is an extremely poisonous compound commonly used in industries such as precious metals mining, metal plating shops, steel mills, and plastic and fertilizer factories. The various industries worldwide produce as much as 1 400 000 tons of toxic * Corresponding author: Email:
[email protected]. 10.1021/ac060890m CCC: $37.00 Published on Web 11/23/2006
© 2007 American Chemical Society
cyanide per year.1 Cyanide’s toxicity results from its propensity to bind to the iron in cytochrome c oxidase, interfering with electron transport and resulting in hypoxia. Since it is lethal to human beings as well as aquatic life, the U.S. EPA regulates cyanide content at the very low levels of 0.2 and 0.005 mg/L for drinking water and environmental primary standards, respectively.2,3 Given its acute toxicity, wide availability in massive amounts, and in light of increasing terrorist activity, there is a pressing need for fast, accurate detection of cyanide at the regulated concentrations. The current detection methods include titrimetric, coulorimetric, voltametric, potentiometric, and spectrophotometric determinations as well as electrochemical, flow injection, ion chromatography, and headspace gas chromatography analyses.4-19 These detection methods suffer from a number of disadvantages such as requiring large sample sizes, long analysis times, high detection limits, and poor precision, largely due to interferences. Current methods are also labor intensive and require significant special skill and training.20 The focus of our current research is to provide a single analytical sensor that combines the advantages found in (1) Sun, H.; Zhang, Y. Y.; Si, S. H.; Zhu, D. R.; Fung, Y. S. Sens. Actuators, B: Chem. 2005, 108, 925-932. (2) http://www.epa.gov/safewater/mcl.html#mcls. (3) EPA, U. Fed. Regist. 1998, 63, FR 44511. (4) Themelis, D. G.; Tzanavaras, P. D. Anal. Chim. Acta 2002, 452, 295-302. (5) Tessier, P. M.; Christesen, S. D.; Ong, K. K.; Clemente, E. M.; Lenhoff, A. M.; Kaler, E. W. Appl. Spectrosc. 2002, 56, 1524-1530. (6) Suzuki, T.; Hiolki, A.; Kurahashi, M. Anal. Chim. Acta 2003, 476, 159165. (7) Safavi, A.; Maleki, N.; Shahbaazi, H. R. Anal. Chim. Acta 2004, 503, 295302. (8) Recalde-Ruiz, D. L.; Andres-Garcia, E.; Diaz-Garcia, M. E. Quim. Anal. 1999, 18, 111-113. (9) Rao, V. K.; Suresh, S. R.; Rao, N. B. S. N.; Rajaram, P. Bull. Electrochem. 1997, 13, 327-329. (10) Presmasiri, W. R.; Clarke, R. H.; Londhe, S.; Womble, M. E. J. Raman Spectrosc. 2001, 32, 919-922. (11) Ng, B. W.; Lenigk, R.; Wong, Y. L.; Wu, X. Z.; Yu, N. T.; Renneberg, R. J. Electrochem. Soc. 2000, 147, 2350-2354. (12) Moriya, F.; Hashimoto, Y. J. Forensic Sci. 2001, 46, 1421-1425. (13) Lu, J. Z.; Qin, W.; Zhang, J. Z.; Feng, M. L.; Wang, Y. Anal. Chim. Acta 1995, 304, 369-373. (14) Licht, S.; Myung, N.; Sun, Y. Anal. Chem. 1996, 68, 954-959. (15) Kelsall, G. H.; Savage, S.; Brandt, D. Electrochem. Soc. 1991, 138, 117124. (16) Jackson, P. E. TrAC-Trend Anal. Chem. 2001, 20, 320-329. (17) Ishii, A.; Seno, H.; Watanabe-Suzuki, K.; Suzuki, O.; Kumazawa, T. Anal. Chem. 1998, 70, 4873-4876. (18) Filipovic-Kovaceic, Z.; Miksaj, M.; Salamon, D. Eur. Food Res. Technol. 2002, 215, 347-352. (19) Curtis, A. J.; Grayless, C. C.; Fall, R. Analyst 2002, 127, 1446-1449. (20) Tarasankar, Pal; Ashes, Ganguly; Durga, S. Maity Anal. Chem. 1986, 58, 1564-1566.
Analytical Chemistry, Vol. 79, No. 1, January 1, 2007 251
many of the previously mentioned cyanide studies.4-19 The toxicity of aquatic cyanides is closely tied to the speciation of the corresponding cyanide compounds.20 The sensor described is especially well suited for monitoring drinking water for protection against accidental or purposeful cyanide contamination, by targeting the most toxic species of aqueous cyanide, particularly free and weak acid dissociable compounds. We propose a new online, real-time method for cyanide detection based on a piezoelectric quartz crystal microbalance (QCM) chemical sensor. Our sensor is based on the piezoelectric effect of a gold-coated crystal and a selective reaction of the crystal’s gold coating with cyanide. The sensor consists of an ATcut quartz crystal sandwiched between two gold electrodes. Placing an electric potential across the electrodes induces the QCM to vibrate at a specific frequency as a result of the piezoelectric effect, which has been described elsewhere.21 This frequency is related to the mass of the gold electrode on the surface of the crystal according to the Saurbrey equation.22 When the gold surface is exposed to a solution containing cyanide, its mass is reduced as the result of the Elsner reaction:23
4Au + 8CN- + 2H2O + O2 a 4Au(CN)2- + 4OH-
(1)
The frequency of the crystal deviates somewhat from the Saurbrey equation as a function of physical properties of the solution.21 However, the difference can be aptly accounted for through calibration of the system. Our laboratory has previously demonstrated that gold-coated QCM can be effectively used as a cyanide sensor in the liquid phase.24 The goal of this work is to present a novel cyanide detection method that has a limit of detection below the EPA water quality criteria and is free of interference from anions commonly found in natural waters including cyanide’s aqueous analogue thiocyanate. This method also offers the advantage of targeting the toxic complexed and free cyanides, since EPA regulates only free cyanides in drinking water.2 Additionally, our technique utilizes a flow system that facilitates continuous online, real-time monitoring with minimal sample preparation. Thus, this technique is ideally suited for automation and allows the analysis to be performed in a manner that is safer, simpler, and more costeffective than previously possible. EXPERIMENTAL SECTION Apparatus. The experimental apparatus is assembled using three commercially available components: a piezoelectric oscillator with a frequency counter, a liquid flow cell that wets only one side of a 10-MHz PZ crystal, and a commercial peristaltic pump. The commercial microprocessor-controlled piezoelectric oscillator and frequency counter (Universal Sensors model PZ105 piezoelectric detector) was used. This is a dual-crystal system with single reference and analytical crystals. The oscillator (21) Bunde, R. L.; Jarvi, E. J.; Rosentreter, J. J. Talanta 1998, 46, 1223-1236. (22) Sauerbrey, G. Z. Phys. 1959, 155, 206. (23) Jungreis, E. J. Chem. 1969, 7, 583-584. (24) Bunde, R. L.; Rosentreter, J. J. Microchem. J. 1993, 47, 148-156. (25) Cyanides in Water. 1994 Annual Book of ASTM Standards; ASTM: Philadelphia, PA, 1994; Vol. 11.2. D-2036-91. D4282-89, pp 79-102.
252
Analytical Chemistry, Vol. 79, No. 1, January 1, 2007
operates on 5-15-MHz crystals while frequency differences range from -32 768 to +32 768 Hz. The system also incorporates automated temperature correction. AT-cut, 10-MHz circular quartz crystals (14-mm diameter) with circular gold electrodes (7.4-mm diameter) are connected to the oscillator using steel pins (1.27mm diameter, 6.35-mm length, 12.34-mm distance between posts) were employed due to their commercial availability and their high electrode surface area. The Sauerbrey equation indicates this sensitivity is a direct result of its relatively large electrode surface area. The crystals are commercially available through a variety of sources, including Universal Sensors (5258 Veterans Blvd., Suite D, Metairie, LA 70006; P.O. Box 736, New Orleans, LA 70148). Solutions were introduced using a Rainin Instrument (Rabbit Model) peristaltic pump. Tygon pump tubings of various sizes were used to obtain flow rates from 0.11 to 1.17 mL/ min. Reagents. All chemicals used were of the highest grade commercially available. Standard cyanide solutions were prepared using 1 g of CN-/L of potassium cyanide stock solution, adjusted to pH 12 with potassium hydroxide for safe handling, unless noted otherwise. All solutions were stored under light- and heatprotected conditions. All anionic solutions used in interference testing were prepared by dilution from their stock solutions of sodium salts. Stock solutions of sodium chloride, sodium nitrate, sodium nitrite, and sodium thiocyanate were diluted in deionized water to a concentration of 100 mg/L and their pH adjusted to 12 by the addition of potassium hydroxide. Digital Signal Processing. The signal was recorded using 1-s increments by means of a PC equipped with the PZ Tools software and a custom-designed Microsoft Excel processing application. Safety Considerations. Due to the toxicity of cyanide, care should be taken to avoid accidental ingestion or skin contact with any solutions containing cyanide. Acidification of cyanide solutions may liberate acutely toxic HCN gas. Complete safety precautions and first aid treatments are aptly described in ref 25. Procedures. The quartz crystal was incorporated into a flow chamber and mounted in the oscillating apparatus/frequency counter. The sample was introduced into the chamber at an angle of 45° to the horizontal of the crystal. The sample flow was continuously maintained in the system using a peristaltic pump. Immediately after crystal incorporation into the flow cell, the system was allowed to equilibrate for 15 min after which the analysis was initiated. The change in mass resulting from the reaction described in eq 1 was recorded. RESULTS AND DISCUSSION Calibration Curve. The detection system presented herein is designed such that it would require minimal human supervision. The simplicity of the operation procedure is comparable to that of a commercially available UV-visible spectrometer. Although according to the literature it was necessary to adjust sample readings by subtracting the average of blank readings before and after due to the instrument drift,24 this correction is no longer necessary as a result of incorporating the crystal into a flow chamber and introducing the sample by means of Tygon tubing controlled with a peristaltic pump. At the same time, the relative
Table 1. Calibration Curve Data sample concn (ppm)
crystal response (Hz/min)
rel std devtn (%)
0.00 0.10 0.20 0.60 1.00 2.00
0.94 4.31 9.23 25.47 40.03 76.74
125.13 22.92 4.99 3.28 5.22 0.56
intercept slope R2
1.43 37.98 0.9992 Figure 1. Crystal response at various pH.
standard deviation of 0.56% shows significant improvement over the previously reported value of 3.7%24 at a cyanide concentration of 2 ppm. Solutions used for calibration were introduced at 1.00 mL/min and were prepared from solid potassium cyanide, with pH adjusted to 12 using potassium hydroxide. Calibration data in Table 1 show a good linearity over a large range of concentrations with the squared linear correlation coefficient of 0.9992. The instrument shows very little drift: within 2 h, the average cyanide concentration recoveries were 101, 103, and 98.8% with the range of 2.7, 3.3, and 5.7%, respectively, for concentrations of 0.20, 0.60, and 1.0 ppm. Detection Limits. Because of only small drift in the crystal’s response, it is possible to increase the detection limits significantly by employing longer analysis times. Standard crystal frequency response is 43 Hz/ppm of cyanide in solution per minute of aspiration. This response is increased to 433 Hz/ppm in 10 min of aspiration and to 2595 Hz/ppm in 60 min of aspiration. The corresponding 10- and 60-min detection limits based on 30 blank readings and three standard deviations (3σ) are 16.1 and 2.7 ppb, respectively. This increase in detection as a function of increased analysis times is a consequence of the direct relationship between dissolution of the crystal’s coating according to the Elsner reaction and the crystal’s vibration frequency. Increased detection times allow for a greater number of cyanide ions to come in contact with the gold electrode, thereby resulting in a greater loss of mass and greater increase in frequency according to the Elsner equation. pH Effects. Cyanide may exist in aqueous solutions in a variety of forms including hydrocyanic acid (HCN) and cyanide ion (CN-). Hydrogen cyanide and free cyanide ion are in equilibrium in aqueous solution according to the following equation:
CN- + H2O a HCN +OHpKa ) 9.36
Consequently, the relative concentration of CN- depends on the pH of the solution. If the proposed cyanide detection system is implemented for low pH conditions, there is a competition for cyanide ions between available protons and gold atoms. It was therefore necessary to investigate the pH effects on detection efficiency. The results of the changes in percentage of signal recovery at various pH levels are shown in Figure 1. Since the instrument detected 5-18% of the signal at expected natural water
Figure 2. Anionic interference determination.
acidity of pH 5-6 this detection system can be used without pH adjustment, when the analysis time is adjusted to yield necessary detection level and reproducibility. The variation of signal with pH also demonstrates that whatever the pH of choice is for the analysis, it has to be kept constant for accurate results to be obtained. Anionic Interferences. A variety of common aquatic anionic species, including thiocyanate, were examined as interfering chemical species. Sulfide ion was not tested, but its interfering effects may be overcome using standard lead precipitation techniques.26 The test samples were composed of 100 ppm interferent solutions spiked with 1 ppm CN- (Figure 2). Of the seven anions tested for possible interfering effects, none showed any noticeable deviation from the spiked cyanide concentration. It has been reported before that chloride ion interferes with cyanide detection.24 It was hypothesized that this interference resulted from the formation of an insoluble complex causing a slight mass gain at concentration as high as 2 ppm. Our data show no such interference, which indicated that no such complex is present at concentrations as low as 1 ppm. The results of interference studies are particularly noteworthy because the weakest point in the current analytical methods for cyanide detection is their susceptibility to interfering ions. Thiocyanate interference is by far the most serious among many chemical species that interfere with cyanide detection. This interference by thiocyanate is partly because thiocyanate is often present in large excess relative to the cyanide content of mining samples (26) U.S. EPA. Methods for Chemical Analysis of Water and Wastes; EPA-600/479-020: U.S. Environmental Protection Agency: Cincinnati, OH, 1983.
Analytical Chemistry, Vol. 79, No. 1, January 1, 2007
253
Table 2. Reproducibility Dependence on Analysis Time
a
Figure 3. Cyanide metal species determination.
and thiocyanate reacts equally as well as cyanide to form chamineT, which is the basis of the colorimetric procedure.27 The U.S. EPA only regulates free cyanide concentrations2 in drinking water. Free cyanide refers to species that may exist under normal environmental conditions as the molecular hydrocyanic acid or the cyanide ion. The rest of the cyanides are found as weak or strong complexes, which are also referred to as complexes amenable to chlorination (CATCs) and complexes not amenable to chlorination, respectively. The method presented herein offers the distinct advantage of detecting free cyanides as well as CATCs, since CATCs can be converted to free cyanides through water purification by chlorination (the supporting data are presented in Figure 3). All anionic cyanide solutions were prepared by dissolving enough solid potassium metal complexes such that total cyanide concentration equaled 2 ppm. The analysis of various metal cyanide complexes of differing formation stability was carried out at the 2 ppm equivalent cyanide concentration and analyzed in triplicates using a 1.063 mL/min flow rate. These data (Figure 3) show no significant recovery of CNATs (Kf > 30) and full recovery of free cyanide and CATs (Kf < 30). This property of the detection system is important in monitoring potable water systems against accidental or deviant cyanide contamination, because the cyanide concentration it yields more closely correlates with the toxicity of the sample, since increased formation constants are associated with decreased bioavailability of cyanide. Sensitivity and Reproducibility. At this point it is important to discuss the factors affecting the sensitivity and reproducibility of the piezoelectric cyanide detection system. Increasing analysis time and optimizing pH resulted in improved sensitivity and has already been discussed above in regard to detection limits and sensitivity. Reproducibility also increases with increased time as noted in the relative standard deviations of the two data sets in Table 2. As predicted by the Saurbrey equation, another important factor is the surface area of the gold electrode in direct contact with analyte solution. Increased surface area will lead to an increased gold mass loss from the crystal’s surface and larger frequency changes, thus improving sensitivity. The sample flow rate was found to be a significant factor as well (Table 3). The sensitivity increase was 4-fold over an 11-fold increase in sample (27) American Public Health Assoc.; American Water Works Assoc.; Water Pollution Control Federation. Standard Methods for the Examination of Water and Wastewater, 18th ed.; American Public Health Association, Washington, DC, 1992; pp 418-434.
254 Analytical Chemistry, Vol. 79, No. 1, January 1, 2007
cyanide concn (ppm)
rel std devna (%)
rel std devnb (%)
0.00 0.10 0.20 0.60 1.00 2.00
-240.75 34.30 5.90 3.47 5.41 0.57
386.13 219.15 67.43 20.72 3.93 9.90
Analysis time is 10 min. b Analysis time is 5 min.
Table 3. Sensitivity Dependence on Flow Rate flow rate (mL/min)
frequency change (Hz)
sensitivity (Hz ppm-1 min-1)
1.171 1.063 0.950 0.840 0.722 0.595 0.457 0.332 0.218 0.109
285.0 253.0 213.0 185.0 174.0 161.0 140.0 100.0 95.0 72.0
57.51 46.24 42.05 34.95 30.34 29.74 26.28 20.03 18.37 14.12
flow rate. Yet, reproducibility decreased as the sample flow rate increased. It was found that the peristaltic pump, used to introduce sample into the flow cell, produced pulsed liquid pressure waves onto the crystal’s surface. This periodically increases the bulk mass sensed by the crystal. Since this pressure variation increases at higher flow rates, the reproducibility of the crystal decreases respectively. Thus, our experimental apparatus limited these studies. Flow rate, as one might expect, improved precision at high flow rates when a more constant solution stream was provided. CONCLUSION In this paper, we present a cyanide detection method suitable for drinking and fresh water, which offers many operational advantages that should allow it to be a widely accepted routine monitoring system. Our system provides reliable, reproducible results over a wide dynamic range of concentrations. The sensor meets and exceeds the U.S. EPA water quality criterion of sensitivity below 5 ppb, while simultaneously maintaining a tolerance for highly concentrated samples containing more than 1000 times that value. It may be operated continuously with a proven instrument drift of nominally 5% over 2 h. The analytical apparatus is simple to operate and does not require special analyst skills. When following the described methodology, little or no sample preparation other than pH adjustment is required. By limiting sample preparation, the possibility of contamination or analyte loss from outgassing is minimized. Results presented using six common aquatic anions show the method is free from interferences by these species. It was shown that even chemically similar species such as thiocyanate, which causes near one-toone positive interference in many analytical methods, have no measurable response. The instrumentation is inexpensive, portable, and provides a tempting solution for on-site analysis.
Moreover, this system has the ability to be incorporated into an automated, inline, real-time monitoring system for both immediate response and continuous monitoring. This system offers a unique opportunity for user customization to various applications. The parameters affecting the performance and sensitivity of the system can be easily adjusted to produce optimal analysis times and reproducibility, depending on individual demands of the desired analysis. Reliable analysis of aqueous cyanide in the ppm concentration range can be obtained in as little as 5 min. Finally, the
system is primarily sensitive to the same aqueous cyanide species that exhibit toxicity to aerobic organisms. This correlation makes the sensor especially attractive for applications aimed at safety and security.
Received for review May 15, 2006. Accepted October 23, 2006. AC060890M
Analytical Chemistry, Vol. 79, No. 1, January 1, 2007
255
