ARTICLE pubs.acs.org/ac
Rugged, Portable Tungsten Coil Atomic Emission Spectrometer Jiyan Gu,† Silvana R. Oliveira,‡ George L. Donati,† Jose Anchieta Gomes Neto,‡ and Bradley T. Jones*,† † ‡
Department of Chemistry, Wake Forest University, Winston-Salem, North Carolina 27109, United States Unesp, Department of Analytical Chemistry, S~ao Paulo State University, P.O. Box 355, 14801-970 Araraquara, SP, Brazil ABSTRACT: Tungsten coil atomic emission spectrometry is an ideal technique for field applications because of its simplicity, low cost, low power requirement, and independence from cooling systems. A new, portable, compact design is reported here. The tungsten coil is extracted from an inexpensive 24 V, 250 W commercial light bulb. The coil is housed in a small, aluminum cell. The emission signal exits from a small aperture in the cell, while the bulk of the blackbody emission from the tungsten coil is blocked. The resulting spectra exhibit extremely low background signals. The atomization cell, a single lens, and a hand-held charge coupled device (CCD) spectrometer are fixed on a 1 6 30 cm ceramic base. The resulting system is robust and easily transported. A programmable, miniature 400 W solid-state constant current power supply controls the temperature of the coil. Fifteen elements are determined with the system (Ba, Cs, Li, Rb, Cr, Sr, Eu, Yb, Mn, Fe, Cu, Mg, V, Al, and Ga). The precision ranges from 4.3% to 8.4% relative standard deviation for repetitive measurements of the same solution. Detection limits are in the 0.04 to 1500 μg/L range. Accuracy is tested using standard reference materials for polluted water, peach leaves, and tomato leaves. For those elements present above the detection limit, recoveries range from 72% to 147%.
M
odern laboratories are capable of performing fast, simultaneous, and multielement analyses for a multitude of sample types and elements.1-3 Transferring this technology to field analysis is not a simple endeavor. A successful field instrument would allow a screening process, so that only the most important samples would be transported back to the central laboratory. This would result in a significant savings of time and effort, while preserving sample integrity. Currently, the two most popular commercially available methods for portable elemental field analyses are X-ray fluorescence (XRF) spectrometry4 and laser induced breakdown spectrometry (LIBS).5 The recent miniaturization of X-ray tubes has resulted in the application of portable XRF in fields such as positive material identification (PMI), mining, and environmental testing. Both XRF and LIBS have their drawbacks. Portable XRF can seldom be employed when analyte concentrations are below ppm levels. The LIBS instrument is often more expensive than XRF. Both techniques suffer from the need for solid standards to produce calibration curves. In addition, tungsten coil atomic absorption spectrometry (WCAAS) has been reported as a portable technique, but it is not commercially available at present.6-10 WCAAS is a technique developed with a more traditional electrothermal atomization device.11-21 WCAAS usually provides LODs (limits of detection) lower than LIBS or XRF, with lower instrumentation costs. The tungsten filament may be extracted from an inexpensive commercially available light bulb. The filament may produce gaseous analyte atoms when heated with a simple solid-state power supply. Emission spectra may be collected with a miniature CCD-based detector and processed with a laptop computer. The whole system may be very small, and no cooling system is required. The power source for the device can be as simple as a 12 r 2011 American Chemical Society
V car battery. Unfortunately, WCAAS is a single element technique, requiring a different hollow cathode lamp for each element determined. Also, sample digestion or extraction methods are necessary for solid samples, since WCAAS requires liquid sample introduction. Tungsten coil atomic emission spectrometry (WCAES) can perform simultaneous multielement determinations without the need for a light source.22,23 With this method, the W-coil is used both as an electrothermal vaporizer and for excitation of emission. Other electrothermal atomic emission devices, such as the graphite furnace, tungsten tube, and molybdenum microtube have been reviewed recently and compared with the W-coil.24 Elements including Al, Co, Cr, Ga, K, Mn, Pb, Rb, Sc, Cr, Ga, In, V, and the lanthanides have been determined by WCAES.25-28 These studies employed a high resolution spectrometer for detection, and this achieved efficient separation of the atomic emission signals from the black blackbody emission arising from the W-coil. Absolute detection limits are at or below the ng/mL level for many elements. The goal of the current work is to render the WCAES small, rugged, and portable. The instrument described here employs a novel aluminum W-coil housing, a single lens, and a miniature CCD spectrometer. Each of these is mounted directly to a ceramic base measuring just 6 cm wide by 30 cm long. No glass or moving parts are included. The device may be operated in the field by powering it with a car battery and controlling it with a laptop computer. A purge gas is supplied by a lecture bottle.
Received: October 22, 2010 Accepted: February 15, 2011 Published: March 03, 2011 2526
dx.doi.org/10.1021/ac1027897 | Anal. Chem. 2011, 83, 2526–2531
Analytical Chemistry
ARTICLE
Figure 2. Photo of the Aluminum Atomization Cell. Figure 1. Photo of the portable WCAES system.
’ EXPERIMENTAL SECTION Instrumentation. Figure 1 is a photograph of the rugged spectrometer. An aluminum tungsten coil housing, an emission signal collection lens, and a small CCD-based spectrometer are mounted on a ceramic rail (1 6 30 cm). The housing is fashioned from a 2.54 cm diameter, 7.5 cm long Al rod (Figure 2). The bottom of the rod is tapped for a 1/4-20 mounting screw. The rod is hollowed out from the top to a depth of 6.5 cm with an internal diameter of 1.75 cm. Near the bottom of the hollow section, a 1/16 NPT port is placed to fit the fitting for the purge gas. The view port for atomic emission is a 1.5 mm hole drilled at a height of 3.3 cm from the base of the housing. A second view port is drilled at a 180° angle to the first one. This allows for easy “line of sight” alignment and also provides a path for atomic absorption measurements. The sample introduction port (a 5 mm hole placed 4 cm from the bottom of the cell) is placed at a 90° angle for the view port. The sample introduction port is not visible in Figure 2. The W-coil is the filament from a 24 V, 250 W light bulb (Osram Xenophot HLX 64655, Augsburg, Germany). The fused silica envelope was removed from the bulb leaving the filament and glass base intact (Figure 2). The base fits in a standard, 2-pronged, ceramic bulb socket. The socket is cemented to the end of an Al mount. The mount is 2.2 cm long. The lower 2 cm of the mount is machined to an outer diameter to match the internal diameter of the housing (1.75 cm). The upper portion of the mount has a diameter of 2.54 cm, so when the mount is inserted in the housing, it acts as a cap. The electrical leads to the ceramic socket are threaded through the mount and sealed with epoxy. The height of the mount and, thus, the position of the coil relative to the view port is adjusted with a machine screw that pushes against the top of the housing. When the height is properly adjusted, the lowest edge of the W-coil is just above the level of the view port. The purge gas (10% H2 in Ar) flows at a rate of 1.0 L/min into the cell and escapes through the view port and sample introduction port. This positive flow prevents air from entering the cell and oxidizing the W-coil. The 2.54 cm diameter, 5 cm focal length fused silica emission collection lens is held with an adjustable X-Y mount. The lens forms a 1:1 image of the view port on the entrance aperture (0.5 mm) of a CCD-based miniature spectrometer (Ocean Optics, Model USB4000, Dunedin FL). The spectrometer is controlled with a laptop computer. Proper optical alignment of the system
ensured that the image of the W-coil was adjacent but not overlapping the entrance aperture. Thus, blackbody radiation from the coil was blocked, while atomic emission originating just below the edge of the filament entered the spectrometer. No significant W emission lines were observed except for a weak W line at 400.8 nm during the final atomization step. Procedure. A 400 W solid-state power supply (Vicor BatMod, Andover, MA) operated by a Visual Basic program provided the heating cycle for the W-coil: 2.7 A, 55 s; 2.4 A, 45 s; 2.2 A, 35 s; 1.5 A, 25 s; 0 A, 10 s; 13 A, 1.5 s; 8 A, 5s. A 70 μL aliquot of solution was deposited directly on the W-coil with an Eppendorf pipet. The heating program started with 2.7 A and slowly dropped to 2.4 A when the majority of the liquid evaporated. This gradually decreasing power cycle prevented the loss of volatile analytes during the drying stage. When the coil was completely dry, the current was adjusted to 0 A for 10 s. This allowed the W-coil to return to room temperature prior to the beginning of the high current atomization step and improved reproducibility. The 13.0 A atomization stage was applied, and the coil reached its full color temperature of 3400 K in approximately 1.5 s. The emission signal was collected during this step. Afterward, a current of 8.0 A was applied for 3 s to remove any sample residue. No memory effects were observed after this cleaning step. Finally, a 30 s cooling step was applied before the next sample injection. The analysis of a single sample required approximately 4 min. For a 15 V, 150 W W-coil, the temperature (T) in Kelvin of the dry coil surface may be estimated from the current (I) in amps: T ≈ 309I þ 325.10 For the 24 V, 250 W W-coil, the highest current available is 13 A, and a current of 13.5 A will melt the W-coil. Given the melting point of W (3680 K), the coil temperature could be approximated with a similar linear relationship T ≈ 250I þ 325. The temperature in the gaseous observation zone is significantly lower than the temperature of the tungsten surface. The temperature measured 2 mm away from the surface is as low as 2200 K.25 While gas phase temperatures were not measured in the current study, this temperature range seems reasonable. The average W-coil lifetime with this heating program was 350 cycles. A properly dried W-coil produced no emission lines for W, even at high currents (below melting). Premature heating of a wet W-coil produced many W emission lines and significantly reduced lifetime. For natural water samples, dissolved carbon may decrease the lifetime of W-coil through the production of tungsten carbides. The emission signal was collected during the high current excitation step. Fifteen successive spectra, each with a 100 ms integration time, were collected during this 1.5 s period. On 2527
dx.doi.org/10.1021/ac1027897 |Anal. Chem. 2011, 83, 2526–2531
Analytical Chemistry
ARTICLE
Figure 3. Simultaneous multielement WCAES determination of Yb, Sr, Ba, Eu, and Cr (1 mg L-1) without background subtraction.
Figure 4. Simultaneous multielement WCAES determination of Al (10 mg L-1), Cr (2 mg L-1), Ga (20 mg L-1), Mn (20 mg L-1), and V (5 mg L-1) without background subtraction.
average, the first emission signals appeared on the eighth spectrum and all signals subsided by the 14th spectrum. As expected, elements with high boiling points (Mn, Cr, Cu) appeared later, and their signals persisted longer than elements with low boiling points (Na, Li). The total integrated signal was calculated by summing the 7th through 14th spectra. Precision for the analysis of identical sample aliquots was in the range of 5.0 to 8.0% in this case. Fifteen test elements were selected to evaluate the new system. These included alkali, alkaline earth, and transition metal elements (spanning the range of melting point, boiling point, and ionization energy). For example, Yb, Sr, Cr, Ba, Eu, and Na were detected simultaneously in one spectrum (400-600 nm,
Figure 3), while Al, V, Cr, Mn, and Ga were detected in another (250-450 nm, Figure 4). Both spectra are the integrated sum of 7 successive spectra as described above. No background signals were subtracted. The limits of detection (LOD) were determined by the IUPAC method: three times the standard deviation in the blank signal divided by the slope of the calibration curve for each element (Table 1). Sample Preparation. All reference solutions were prepared from dilution of single element stock solutions (1000 mg L-1, SPEX CerPrep, Metuchen, NJ, USA) with distilled-deionized water (Milli-Q, Millipore Corp., Bedford, MA, USA). One water standard reference material (National Institute of Standards and Technology, SRM #1643e, Gaithersburg, MD, USA), one 2528
dx.doi.org/10.1021/ac1027897 |Anal. Chem. 2011, 83, 2526–2531
Analytical Chemistry
ARTICLE
Table 1. Analytical Figures of Merit for Portable WCAES and Reference Materials references element
a
wavelength (nm)
LOD (μg L-1)
RSD (%)
recovery
certified
accuracy
Ba
553.5
0.72
7.1
585 ( 44
544.2 ( 5.8
Cs
852.1
0.47
6.0
22.7 ( 1.7
20b
114%
Li
670.8
0.04
1.5
18.2 ( 0.8
17.4 ( 1.7a
105%
Rb
780.0
0.2
2.6
32.6 ( 4.7
14.14 ( 0.18a
231%
Cr
520.8
2.3
8.4
121 ( 35
100c
121%
Sr
460.7
0.22
3.6
474 ( 33
323.1 ( 3.6a
147%
Eu
459.4
8.1
8.0
89 ( 21
Yb Mn
398.8 403.1
6.9 50
6.0 6.7
70.3 ( 7.7
98 ( 3d
72%
Mn
403.1
50
6.7
249 ( 67
246 ( 8e
101%
Fe
385.9
400
5.3
218 ( 26
218 ( 14d
100%
Fe
386.9
400
5.3
360 ( 47
368 ( 7e
98%
Cu
324.8
300
6.4
Mg
285.2
1500
6.5
0.432 ( 0.008
0.455 ( 0.034d
95%
V
318.4
150
5.2
238 ( 23
250c
95%
V Al
386.4 397.1
180 67
5.0 7.0
225 ( 15 527 ( 63
250c 500c
90% 105%
Al
394.4
53
8.3
495 ( 54
500c
99%
Ga
404.4
200
5.0
10.2 ( 0.5
10b
102%
Ga
418.6
160
9.5 ( 1.7
10b
95%
-1
b
4.6 -1
-1
100b
108%
89%
NIST SRM 1643e (μg L ). Tap: Sample is spiked in tap water (mg L ). VHG (mg L ). NIST 1547 peach leaves (μg/g). e NIST 1573a tomato leaves (mg/kg).
polluted water standard reference material (Water Pollution Standard 1, Product Number WPS1-100, VHG Laboratories Inc., Manchester, NH, USA), and two solid samples from the National Institute of Standards and Technology (peach leaves #1547 and tomato leaves #1573a) were used to check the accuracy of the instrumentation. The water samples were diluted with distilled-deionized water into the linear range of the calibration curves. The solids were digested using concentrated HNO3 and 30% H2O2 (v/v). Approximately 0.25 g was weighed accurately into a plastic extraction container, and a 2 mL aliquot of acid was added. The mixture was left to react for 1 min, and then, 3 mL of H2O2 plus 2 mL of deionized water were added. The plastic container was placed in an aluminum hot block at 100 °C, and the extraction was carried out for 2 h. To prevent the sample from drying, 2 aliquots (1 mL each) of deionized water were added to the sample at different intervals during the digestion procedure. The digested samples were allowed to cool for 20 min and subjected to filtration through coarse filter paper (Fisherbrand). The filtrate was diluted to a total of 25 mL with deionized water. This procedure was carried out in triplicate. Safety Considerations. Material safety data sheets were consulted before using each chemical reagent. Essential safety precautions (gloves, glasses, etc.) were taken in each step of the analysis. Aqueous waste was stored in glass containers prior to disposal.
’ RESULTS AND DISCUSSION Performance Characteristics. The Al atomization cell (Figure 2) is more durable, smaller, and simpler in design than the glass atomization cells reported previously.7,8 Having the 3 components (cell, lens, spectrometer) all fixed permanently to a
c
a
d
simple rail improves portability. In addition, the 24 V, 250 W W-coil is larger in size than those reported previously, and it is capable of holding a significantly higher sample volume (75 μL). The associated 400 W power supply may apply up to 13 A to the coil, enabling complete data collection in a very short time (1.5 s). The blackbody emission from the W-coil has reached its peak by this point, indicating that the maximum temperature has been achieved. The very short atomization period improves the lifetime of the W-coil, and the rapid heating minimizes analyte loss from the emission observation area prior to full temperature. For most W-coil devices, a major concern involves the background emission from the filament at high temperature. The filament, after all, is intended to be used as a light source. Previous WCAES devices relied upon correction of this background by subtracting a blank signal collected during a similar atomization cycle. While this approach is effective, the signal-to-noise ratio suffers, since the noise on the background signal is not subtracted. The Al cell is designed with two background blocking features. First, the view port on the cell has a very small diameter. The W-coil is positioned so that the edge of the filament is slightly above the port. Most of the blackbody emission, therefore, cannot escape from the view port (at least not in the direction of the spectrometer). On the spectrometer end of the optical rail, a second aperture further eliminates blackbody emission. This aperture is created by drilling a 0.5 mm hole in the center of a cap threaded to fit the SMA fiber optic entrance port on the spectrometer. The lens is adjusted so that the image of the filament (resulting from light scattered through the view port) does not pass through the detector aperture. With this design, no background spectra are subtracted even in the visible region (Figure 3). The signal intensity of each peak is calculated 2529
dx.doi.org/10.1021/ac1027897 |Anal. Chem. 2011, 83, 2526–2531
Analytical Chemistry by subtracting the average of two nearby flat baseline points from the peak height. Analytical Figures of Merit. Analytical figures of merit are reported in Table 1. The blank noise used to calculate the LOD was taken as the standard deviation calculated for 17 blank analyses. WCAES LODs are very low (sub ng mL-1) for elements having strong emission lines in the visible region of the spectrum (Ba, Cs, Li, Rb, and Sr). Moderate LODs (2-200 ng mL-1) were observed for elements with emission lines near 400 nm (Cr, Eu, Yb, Mn, V, Al, and Ga). In the UV region, the spectrometer falls off in sensitivity and higher LODs result (Fe, Cu and Mg). One indication that the spectrometer is not as sensitive in the UV and near-UV region is the result for Pb. The primary emission line for Pb at 405.8 nm had a 500 ng mL-1 detection limit using a high resolution spectrometer with a cooled CCD spectrometer.23 This line is not observed with the current system. The linear dynamic range (LDR) for most elements is in the range of 1-2.5 orders of magnitude. Elements with lower LODs tend to have the larger LDR. Calibration curvature was observed for all elements in the concentration range from 5 to 50 μg/mL. This could be due to self-absorption. During the atomization step, the analyte vapor cools with distance from the coil surface. Ground-state atoms will, therefore, exist between the coil and detector. The precision for the 15 test elements ranged from 4% to 8% RSD, using concentrations 50 times greater than the detection limit. Analysis of the reference materials was performed by the calibration curve method for the peach leaves, tomato leaves, and tap water. The standard addition method, as described previously,28 was required for the two water samples containing very high concentrations of many elements (polluted water and NIST SRM 1643). All elements (except for Cu and Yb) were detected in the reference samples with high recovery (Table 1). For Cu, the LOD was too high for accurate analyses, and Yb (398.8 nm) had a significant spectral overlap from large amounts of Al (397.1 nm) using the low resolution portable spectrometer. Interferences. The W-coil is most effective for the excitation of elements with relatively low excitation energy. Compared to emission sources such as arcs, sparks, and plasmas, much less spectral interference is observed with the W-coil. In most cases, therefore, the low resolution spectrometer was adequate. Some isolated interference cases were observed, however. For example, the Yb line at 398.8 nm experienced interference from an Al line at 397.1 nm when Al was in high concentration. Also, Fe produced a large number of WCAES lines in the 340-380 nm region. At Fe concentrations above 100 μg mL-1 in the sample, weak analyte emission lines in this region could be completely obscured. Alternative wavelengths should be employed in these cases. Table 1 lists the primary WCAES emission wavelengths, but many others were observed such as Li 610.3 nm, 323.2 nm; Sc 391.2 nm; Ti 399.8 nm, 430.5 nm, 498.1 nm; Cr 357.8 nm, 425.4 nm; Ni 341.4 nm, 344.6 nm; Cu 327.3 nm; Zn 330.2 nm, 334.5 nm; Rb 794.7 nm; Ag 328.0 nm; Cd 346.6 nm; Cs 894.3 nm; and Pt 330.1 nm, 340.8 nm. For the most part, the major emission lines reported for graphite furnace atomic emission spectrometry (GFAES) can also be detected by WCAES for these elements.24 Matrix effects were not investigated in detail, and they were minimized by sample dilution or by employing the method of standard additions. WCAES atomization temperatures are similar to those for GFAAS, but the W-coil is an open atomizer; thus, the sample matrix is more dilute during the atomization step. When matrix effects persist, preconcentration methods could be used to separate matrix elements from the sample. The high
ARTICLE
current attained with the WCAES system (13 A) may provide a more robust emission environment, compared with previously reported WCAES devices. For example, Sr(II) and Ba(II) lines are observed in this high temperature environment (Figure 3).
’ CONCLUSION WCAES is one of simplest methods for the simultaneous multielement analysis of liquid samples. The rugged design reported here has been evaluated with 15 test elements. Further improvement in the UV region could be achieved with a detector that is more sensitive in this region. The accuracy of the potentially portable device is high, especially at higher wavelengths. Relatively simple and potentially portable sample preparation procedures are effective. Internal standard or standard addition methods may be necessary, especially when analytes are present in low concentrations while matrix elements are abundant. The portable WCAES device should be a very effective screening tool for field analyses. ’ AUTHOR INFORMATION Corresponding Author
*E-mail:
[email protected].
’ ACKNOWLEDGMENT This material is based upon work supported by the National Science Foundation and the Department of Homeland Security through the joint Academic Research Initiatives program: NSF CBET 0736214 and DHS 2008-DN-077-ARI020-03. ’ REFERENCES (1) Jackson, K. W.; Lu, S. J. Anal. Chem. 1998, 70, 363r–383r. (2) Horlick, G. Anal. Chem. 1980, 52, R290–R305. (3) Beauchemin, D. Anal. Chem. 2008, 80, 4455–4486. (4) Hou, X. D.; He, Y. H.; Jones, B. T. Appl. Spectrosc. Rev. 2004, 39, 1–25. (5) Cremers, D. A.; Chinni, R. C. Appl. Spectrosc. Rev. 2009, 44, 457–506. (6) Hou, X. D.; Jones, B. T. Microchem. J. 2000, 66, 115–145. (7) Batchelor, J. D.; Thomas, S. E.; Jones, B. T. Appl. Spectrosc. 1998, 52, 1086–1091. (8) Sanford, C. L.; Thomas, S. E.; Jones, B. T. Appl. Spectrosc. 1996, 50, 174–181. (9) Bruhn, C. G.; Ambiado, F. E.; Cid, H. J.; Woerner, R.; Tapia, J.; Garcia, R. Anal. Chim. Acta 1995, 306, 183–192. (10) Salido, A.; Jones, B. T. Talanta 1999, 50, 649–659. (11) Ezer, M.; Elwood, S. A.; Jones, B. T.; Simeonsson, J. B. Anal. Chim. Acta 2006, 571, 136–141. (12) Hayashi, H.; Hara, Y.; Tanaka, T.; Hiraide, M. Bunseki Kagaku 2001, 50, 631–634. (13) Hou, X. D.; Yang, Z.; Jones, B. T. Spectrochim. Acta, Part B 2001, 56, 203–214. (14) Nobrega, J. A.; Rust, J.; Calloway, C. P.; Jones, B. T. J. Braz. Chem. Soc. 2005, 16, 639–642. (15) Williams, M.; Piepmeier, E. H. Anal. Chem. 1972, 44, 1342–1344. (16) Suzuki, M.; Ohta, K. Talanta 1981, 28, 177–181. (17) Suzuki, M.; Ohta, K.; Yamakita, T.; Katsuno, T. Spectrochim. Acta, Part B 1981, 36, 679–686. (18) Suzuki, M.; Ohta, K.; Yamakita, T. Anal. Chim. Acta 1981, 5, 209–213. (19) Krenzelok, M; Rychlovsky, P; Volny, M; Matousek, J Analyst 2003, 128, 293–300. 2530
dx.doi.org/10.1021/ac1027897 |Anal. Chem. 2011, 83, 2526–2531
Analytical Chemistry
ARTICLE
(20) Silva, M. M.; Silva, R. B.; Krug, F. J.; Nobrega, J. A.; Berndt, H. J. Anal. At. Spectrom. 1994, 9, 861–865. (21) Zeng, C.; Wen, X.; Tan, Z.; Cai, P.; Hou, X. Microchem. J. 2010, 96, 238–242. (22) Rust, J. A.; Nobrega, J. A.; Calloway, C. P.; Jones, B. T. Spectrochim. Acta, Part B 2005, 60, 589–598. (23) Rust, J. A.; Nobrega, J. A.; Calloway, C. P.; Jones, B. T. Spectrochim. Acta, Part B 2006, 61, 225–229. (24) Rust, J. A.; Donati, G. L.; Afonso, M. T.; Nobrega, J. A.; Jones, B. T. Spectrochim. Acta, Part B 2009, 64, 191–198. (25) Donati, G. L.; Calloway, C. P.; Jones, B. T. J. Anal. At. Spectrom. 2009, 24, 1105–1110. (26) Donati, G. L.; Gu, J.; Nobrega, J. A.; Calloway, C. P.; Jones, B. T. J. Anal. At. Spectrom. 2008, 23, 361–366. (27) Donati, G. L.; Kron, B. E.; Jones, B. T. Spectrochim. Acta, Part B 2009, 64, 559–564. (28) Donati, G. L.; Gonzales, M. H.; Nobrega, J. A.; Jones, B. T. Anal. Lett. 2010, 43, 1723–1733.
2531
dx.doi.org/10.1021/ac1027897 |Anal. Chem. 2011, 83, 2526–2531
