ARTICLE pubs.acs.org/ac
Determination of Trace Perchlorate Concentrations by Anion-Selective Membranes and Total Reflection X-ray Fluorescence Analysis Vasilios S. Hatzistavros and Nikolaos G. Kallithrakas-Kontos* Analytical and Environmental Chemistry Laboratory, University Campus, Technical University of Crete, GR-73100 Chania, Greece
ABSTRACT: In the present work a method for the determination of perchlorate trace levels by total reflection X-ray fluorescence (TXRF) is introduced. Perchlorate anions were concentrated on anion-selective membranes that had been prepared on the surface of TXRF quartz reflectors. Various complexation substances were used in the membranes. The reflectors were immersed in water solutions containing nanogram per milliliter (ppb) concentrations of perchlorate. After this step, the reflectors were taken out of the solution and they were analyzed by TXRF, using a copper X-ray tube and helium flow on the target (to lower the argon peak which is present in the air). The effects of various experimental parameters were examined, and the possibility of discrimination between chloride and perchlorate anions was shown. Minimum detection limits lower than 1 ng/mL and good linearity at the concentration range of 150 ng/mL were achieved. The method is applicable for the analysis of perchlorate in drinking water samples.
P
erchlorate originates as contaminant in the environment mostly from its use as an oxidizer in solid propellants for rockets, but it can be also found in fireworks, airbags, fertilizers, etc.1 Perchlorate is highly soluble in water, so it is easily transferred to the ground and surface water and also appears to soil and food. It has a high accumulation rate in plants, and it cannot be removed with the usual water treatments. The widespread presence of perchlorate in water, foods, and plants leads to perchlorate exposure of humans and wildlife.24 Due to its properties perchlorate has been characterized as an emerging contaminant.5,6 Thyroid uses iodine for hormone production. Perchlorate has a similar size and properties with iodine and may block normal thyroid function (hypothyroidism) for a prolonged time exposition, due to its chemical similarity with iodine; high perchlorate concentrations interfere with iodine and block its absorption in the thyroid (hypothyroidism).710 Very little is known about the health effects of low-level long-term exposure to perchlorate, such as occurs with contaminated drinking water, although it is known that perchlorate shows thyreostatic activity by inhibiting iodine uptake and may therefore hinder the ability of humans to produce hormones and regulate their metabolism. Perchlorate salts have been used to treat patients with hyperactive thyroid glands (Graves’s disease) and to carry out diagnostic tests. There has been much debate over what level of perchlorate is safe for humans to consume daily. After considering the recommendations of the National Academy of Science National Research Council,11 the Environmental Protection Agency set a perchlorate reference dose for adult humans of 0.7 μg/ (kg 3 day) which translates to a drinking water equivalent level (DWEL) of 24.5 μg/L.12 The DWEL is based on a 70 kg adult that consumes 2 L of water per day and assumes 80% of the perchlorate exposure will be from drinking water. Since perchlorate has been detected in a variety of non-water samples, r 2011 American Chemical Society
other food sources should also be taken into account when assessing exposure and risk. The determination of trace or ultratrace perchlorate ions has been carried out directly or indirectly by a variety of analytical methods including spectrophotometry,13 atomic absorption spectrometry,14 attenuated total reflectance and chromatography.15 The EPA method for the determination of perchlorate in drinking water16 is using two-dimensional ion chromatography with suppressed conductivity. The main advantage is that initial sample loading on a 4 mm column gives the possibility of large injection volumes. The two-dimensional separation provides signal enhancement of the perchlorate; major disadvantages are interferences caused by contaminants in the reagents and sample-processing apparatus as well as in the analyzed sample matrixes. However, most of these methods are either timeconsuming or need expensive instrumentation, well-controlled experimental conditions, and most importantly suffer from various interferences of cationic or anionic species. Hence, major efforts have been made to develop more convenient direct methods for the easy and inexpensive assay of perchlorate ions in different samples. Ionselective polymer membrane electrodes incorporating ion carriers with unique characteristics, such as easy preparation, fast response, low cost, wide linear range, relatively low detection limit, and especially good selectivity, can be very suitable tools for the determination of perchlorate ions in different samples.1719 The anion-selective membranes, suggested in the present work, are easily prepared and show great resistance, high selectivity over the desired ions, and physical and chemical stability. There is no need of preconcentration, the analysis time is very small (100 s), and the experimental steps followed are simple. On the other hand, this type of anion-selective membrane Received: December 20, 2010 Accepted: March 21, 2011 Published: April 04, 2011 3386
dx.doi.org/10.1021/ac103295a | Anal. Chem. 2011, 83, 3386–3391
Analytical Chemistry cannot be effective in very complex matrixes like waste waters. Additionally, the equilibration time (between membrane and water sample) can be from 2 to 20 h, depending on desired detection limit and irradiation time. The traditional anion-selective membrane electrodes based on ion exchangers always display a Hofmeister selectivity sequence, as follows: perchlorate > thiocyanate > iodide salicylate > nitrate > bromide > nitrite > chloride > sulfate, in which the membrane selectivity is controlled by the free energy of hydration of the ions involved; recent studies have been more concentrated on the electrodes using plasticized poly(vinyl chloride) (PVC) membranes incorporating different complexing reagents as carriers.20 In a previous work21 a new technique had been developed by fabrication of suitable membranes on the surface of total reflection X-ray fluorescence (TXRF) reflectors. These membranes selectively collected bromate anions and result in lower detection limits in TXRF analysis; they were made from PVC matrix with Aliquat-336 extractant. In the present study novel membranes, suitable for perchlorate anion collection, were prepared to detect low perchlorate concentrations in water solution (drinking water, high-purity water). The membranes were deposited on the center of the quartz reflectors, and they contained different complexing reagents as well as auxiliary compounds. The reflectors with the membrane were immersed in water solutions containing perchlorate concentrations. The system (reflectors plus membrane plus water solution) was left to equilibrate. After this period the reflectors were taken out of the solutions, and they were analyzed by TXRF. Removal of the membrane from the quartz (after analysis) was simple, and the amount of the reagents used was very small, since the quantities of the deposited membranes were only a few micrograms. The anion membrane type that demonstrated the best performance for the detection of perchlorate ions was composed from ethyl vinyl acetate (EVA) with tricapryl methyl ammonium chloride (Aliquat-336). Aliquat-336 is a quaternary ammonium salt that can be used as extractant for anion exchangers. Different complexing agents were introduced in the EVA membranes for perchlorate collection. Among them phenolphthalein proved to be the best. The main advantages of TXRF are its ability to detect exceptional small quantities (nanograms or picograms), to analyze simultaneously most of the elements (with atomic number higher than of magnesium), and low cost per sample analysis. The acquisition time depends on the sample concentration; usually 100 s is enough, but longer times can also be used.
’ EXPERIMENTAL SECTION Membrane Composition. A variety of membrane matrixes were tested by mixing the components that are given in Table 1. The different matrixes were chosen because of their ability to complex anions easily from the solutions. Complexing Reagent. The ligand solutions were prepared by the dissolution of 3 mg of complexing reagent to 2 mL of high-purity water (type I-ASTM-D1193-91, resistivity 18.0 MΩ 3 cm). The tested ligands were the following: L1, 1-nitroso-2-naphthol (Fluka Chemika no. 73910); L2, 4-nitrocatechol (Aldrich 17960); L3, morin hydrate (Fluka Chemika no. 69870); L4, antipyrine (Aldrich 200-486-6); L5, phenyl acetate (Aldrich 108723); L6, o-dianisidin (Fluka Chemika no. 33430); L7, citric acid (MCW 0627); L8, thiourea (Riedel de-Haen no. 33717); L9, triethylenetetramine-N,N,N0 ,N00 ,N000 ,N000 hexaacetic acid (Fluka Chemika no. 90471); L10, dibenzoylmethane (Fluka Chemika no. 33570); L11, dithizone (Fluka Chemika no. 43820); L12, ammonium pyrrolidinedithiocarbamate (Fluka
ARTICLE
Table 1. Membranes of Various Matrix Compositions That Were Tested for Perchlorate Selection Abilitya no.
matrixb
Aliquat-336
solventb
M1 M2 M3
EVA (60%)
40
THF
EVA (70%) EVA (40%)
30 60
THF THF
M4
EVA (50%)
M5
CTA (57%)
14
50
THF
29
M6
CTA (28%)
THF
62
10
M7
CTA (50%)
THF
10
40
THF
plasticizer (dibutyl phthalate)
a
Tetrahydrofuran was used as solvent in all cases. b EVA = poly(ethylene-co-vinyl acetate); CTA = cellulose triacetate; THF = tetrahydrofuran. Chemika no. 09935); L13, 2-mercaptobenzothiazole (Fluka Chemika no. 63720); L14, 4(2-pyridylazo)resorcinol (Fluka Chemika no. 82970); L15, rhodamine B (Fluka Chemika no. 83690). Organic Dyes. The organic dyes solutions were prepared by the dissolution of 3 mg of the reagent to 2 mL of organic solvent. The tested dyes (usually used as indicators) were the following: I1, phenolphthalein (Fluka Chemika 34605); I2, thymol blue (Fluka Chemika 32728); I3, methyl orange (Fluka Chemika 32624); I4, bromothymol blue (Fluka Chemika 32822); I5, eriochrome black-T (Fluka Chemika 32751). Membrane Preparation. The membranes were prepared directly on the center of the quartz reflector by mixing 6 μL of the membrane solution and 6 μL of the ligand solution. First, the ligand solution was placed on the quartz, and then the membrane solution was added on the ligand solution. The resulting solution was dried with the use of an IR lamp. Oven drying did not cause any difference in the membrane adhesion or in the bromate analysis. As the membrane solution is not hydrophilic there was no need for quartz siliconization of the slides, for production of a hydrophobic surface. Preparation of the membrane spot was similar to the classical TXRF way for the preparation of sample spots on quartz carriers.22 The produced membranes on the quartz surfaces were analyzed with TXRF in order to determine any chlorine blank traces. Membrane removal from the quartz was quite easy; quartz with membrane was immersed in ethanol solution for 30 min, and then the membrane was easily removed from the surface of the quartz (itself or with the aid of a soft paper). No memory effects (like chloride detection) were observed after the analysis of the cleaned quartz. Chloride Excess Removal. The selectivity of the membrane toward chloride ions was negligible for small concentrations (lower than 1 ppm); on the other hand the presence of a large excess (1015 ppm) of chloride ions in the drinking water affects the performance of the membrane toward perchlorate ions selectivity and their discrimination from chlorides ones. In order to avoid the problem of chloride anions excess, drinking water samples were spiked with perchlorate anions of known concentration and silver nitrate was added. The solution was stirred and heated until boiling; after removal of the of silver chloride precipitate the solution was analyzed by TXRF. No chloride interference to perchlorate analysis and no perchlorate loss were identified under this process. TXRF Analysis. For analysis of the samples, the TXRF method was used. The X-ray total reflection was performed by a TXRF 3387
dx.doi.org/10.1021/ac103295a |Anal. Chem. 2011, 83, 3386–3391
Analytical Chemistry
ARTICLE
Figure 1. Membrane matrix comparison toward perchlorate X-ray yield. The composition of every matrix is given in Table 1.
module designed and produced at the Vienna Atominstitut. The experimental conditions were the following: voltage 30 kV, current 20 mA, fine-focus copper X-ray tube (Seifert FK 60-04 AG), a Seifert 150-debyeflex 3000 high-voltage generator, and helium flow on the target (to lower the argon peak as argon is present in the atmospheric air). The produced X-rays were detected by an Oxford Si (Li) detector with an 80 mm2 surface area and a resolution of 155 eV at 5.9 keV. The produced signal was amplified by a Tennelec TC-244 spectroscopy amplifier and registered with a personal computer multichannel analyzer card (PCA-II Nucleus). Analysis time was in most cases equal to 100 s. Peak integration and background subtraction were performed by the computer program AXIL distributed by the International Atomic Energy Agency (IAEA).23 pH Effect. The produced membranes were put in different pH solutions (2.00, 5.00, 7.00, and 10.00). The solutions contained 50 ng/mL perchlorate anions, and they were acidified with H2SO4; the pH = 10 was prepared with the use of NaOH. Linearity. The linearity of X-ray perchlorate yield was examined in high-purity and in drinking water as well. The concentrations in both types of water were 3, 5, 10, 25, and 50 ng/mL, and the membrane equilibration time was 1 day. The results show a very good linearity for both cases, and it can be used for the quantification of the perchlorate content in water samples. The minimum detection limit (MDL) was estimated to be 3 times the background standard deviation (square root of the background). For 50 mL of sample and 100 s of live time it was found to be equal to 0.8 ng/mL for high-purity water and 0.9 ng/mL for drinking water. The detection limit can be improved further by increasing the counting time. When the calibration lines are compared, it is obvious that the linearity experiments gave similar results, especially in the low part-per-billion level (050 ng/mL). In order to control any statistical fluctuations, during the calibration procedure, each sample (calibration point) was analyzed in triplicate. The relative standard deviation (RSD) was determined to be equal to 1%.
’ RESULTS AND DISCUSSION In order to find the optimal matrix membrane composition, membranes of various matrixes and different compositions (ration in matrix/plasticizer, extractant) were put in 50 mL of high-purity water solution containing 50 ng/mL of perchlorate ions. The membranes were prepared by physical immobilization of the extractant in a plasticized polymer matrix. The properties of these membranes can be regulated by appropriate selection of the matrix forming polymer, plasticizer, and extractant. The membranes were classified according to the chlorine X-ray yield, from the highest to the lowest. The composition of the tested matrixes is presented in Table 1. No PVC matrixes were examined, because chloride is a
basic constituent of the PVC. Alternatively, two different types of membrane matrixes were used: cellulose triacetate (CTA) (with plasticizer) and EVA (without plasticizer). The best performance was taken from membrane which was created by mixing 60% w/w EVA, 40% w/w Aliquat-336 and 5 mL of tetrahydrofuran (THF) as shown in Figure 1. EVA shows great plasticity of high mechanical resistance, and due to its low transition temperature, it can be used without any plasticizer. The plasticizer used with the CTA membrane was dibutyl phthalate, which is the most commonly used plasticizer. It belongs to the phthalate esters group that are by far the most successive plasticizers, primarily to make soft and flexible membranes (they produce long polymer molecules to slide against one another).The presence of plasticizer is very important for selective membranes as it improves the sensitivity and selectivity of the sensor with a dual function: Besides modifying the distribution constant of the used ionophore, it also acts as a plasticizing agent by enabling the homogeneous solubilization of the membrane ionophore. On the contrary the lost by leaching out, diffusion, and evaporation of the plasticizer prejudices the analytical performance of the membrane. Aliquat-336 except for its extraction properties acts as a plasticizer as well as its molecules are entangled with the polymer chain penetrating between polymer strands and essentially neutralize their polar groups with its own polar groups and/or increasing the distance between the polymer strands and reducing the strength of intermolecular forces acting between them. As an anion receptor it uses binding forces (electrostatic interactions, hydrogen bonds, and covalent coordination) in order to complex with perchlorate anions. The results for the best complexing ligands in combination with the best membrane (matrix no. 1) are presented in Figures 2 and 3. Fifteen different complexing agents (Figure 2) and five different organic dyes (Figure 3) were examined with the same membrane, and the best yield was taken with phenolphthalein (organic dye no. 1) for perchlorate detection. The complexing ligands and organic dyes were chosen in order to form complexes with the analyzed ions and to transfer them into the membrane matrix. All these complexing agents contain different functional groups (like carboxyls, phenols, thiols, and amines) that form strong complexes. Phenolphthalein, which had the best performance, contains two phenolic hydroxyl groups that can covalently immobilize it to the EVA membrane, by ionpair extraction; phenolic compounds (as hydrogen-bonding donors) enhance the extraction of perchlorate on the membrane. The pH experiments are presented in Table 2. The results showed that analysis was independent of the pH of the solution in the range of 28. Changes are observed at higher pH values, where the yield decreased due to competition from the hydroxide ions (the hydroxide ion competes with the perchlorate one for the cationic site in the membrane). At lower pH values, membranes have poor response due to decomposition and instability effects. 3388
dx.doi.org/10.1021/ac103295a |Anal. Chem. 2011, 83, 3386–3391
Analytical Chemistry
ARTICLE
Figure 2. Comparison of different ligands toward perchlorate X-ray yield. The composition of every ligand is given in the text.
Figure 3. Comparison of different organic dyes toward perchlorate X-ray yield. The compositions of the dyes are given in the text.
Table 2. Chloritne X-ray Yield As a Function of Different pH Values (Perchlorate Concentrations 50 ng/mL) pH
perchlorate yield
2.0
6.27 ( 0.55
5.0
6.31 ( 0.53
7.0
6.45 ( 0.56
10.0
4.83 ( 0.48
The presence of argon in the atmospheric air is a problem, because the argon Ka X-rays are near to the chlorine Ka (2.957 for argon and 2.622 for chlorine). This problem was overcome by supplying a helium flux at the target, resulting in a lowering of the argon peak. Figure 4 presents the two different spectra, with and without the helium supply; it is obvious that the argon peak, which overlaps with the Kb chlorine peak, has been almost moved out (in the spectrum in gray color) after putting the helium flux on the target. Figure 5 presents the performance of the membrane toward chloride and perchlorate ions in the same concentration. In very low concentrations of chloride ions (1100 ppb), the membrane showed no complexation effect to chloride ions contrary to perchlorate complexation, verifying the selectivity of the membrane toward perchlorate ions and following the Hoffmeisterbased selectivities mentioned in the introduction. In order to succeed in perchlorate and chloride discrimination, silver nitrate was added in a solution containing chloride and perchlorate ions; the precipitation of chloride ions (as silver chloride) did not affect the perchlorate ones. Figure 6 presents the effect of the addition of silver nitrate in a drinking water solution which had a 10 mg/kg chloride concentration. The remaining
chlorine blank signal is due to the use of the EVA membrane, which includes traces of chloride in its original structure (coming from the Aliquat-336). The chloride blank value was taken into account in all other measurements; no significant blank alterations were observed in analyses before and after equilibration in pure water. The chlorine Ka X-ray line does not interfere with elements normally present in drinking water (in the form of NO3, NH4þ, Naþ, Mg2þ, SO42, Kþ, Ca2þ, Mn2þ, Fe2þ, etc.) as well as with the used complexing reagents (they contain H, C, N, S). The complexation time of perchlorate anions to the membrane was also examined, and the results are presented in Figure 7. Eighteen hours after their residence in the solution perchlorate ions are collected on the membrane in high yield. Stirring and mild heating in order to enhance the mobility of the anions in the solutions and to accelerate the complexation on the membrane had no better impact. Even though the many hours needed for the complexation of perchlorates on the membrane is a drawback if samples must be analyzed immediately, in this case smaller equilibration times (e.g., 2 h) with longer X-ray irradiation time can achieve the same minimum detection limits as well. Concerning the method of quantification an internal standard is more useful when many analytes are determined and supposes that the added substance shows the same behavior as the analyte. In the present case, a single-element calibration curve or standard addition method (to eliminate matrix effect) can be used. Linearity experiments (Figure 8) gave similar results for both ultrapure and drinking water especially in the low part-per-billion levels (050 ng/mL) as the matrix in the analyzed samples was the same. So no standard addition calibration is needed, except in the high perchlorate concentrations (higher than 50 ng/mL). Acceptable precision was obtained, as the RSD % of the method 3389
dx.doi.org/10.1021/ac103295a |Anal. Chem. 2011, 83, 3386–3391
Analytical Chemistry
ARTICLE
Figure 4. Comparison of two X-ray spectra: the gray line represents the spectrum with helium flux, whereas the black one is without.
Figure 5. Comparison of membrane M1 performance toward chloride and perchlorate anions.
Figure 6. Perchlorate X-ray yield after removal of chloride ions from drinking water with the addition of silver nitrate and precipitation of chloride as silver chloride.
Figure 7. Perchlorate X-ray yield as a function of the membrane equilibration time in a 50 ng/mL solution. 3390
dx.doi.org/10.1021/ac103295a |Anal. Chem. 2011, 83, 3386–3391
Analytical Chemistry
ARTICLE
Figure 8. Linearity of the perchlorate X-ray yields as a function of its concentration in high-purity water (R2 = 0.998) and in drinking water (R2 = 0.997).
was 1% and the correlation coefficients obtained for calibration curves ranging from 1 to 50 ng/mL were 0.998 for high-purity water and 0.997 for drinking water, respectively.
’ CONCLUSIONS The use of membrane complexation after the precipitation of silver chloride showed that the detection of perchlorate in drinking water can be achieved at the low nanogram per milliliter concentration with TXRF analysis. The results can be concluded as following. (1) The proposed liquid perchlorate-selective membranes could be used for perchlorate analysis. (2) The best membrane was based on the EVA plus Aliquat-336 plus phenolphthalein system. (3) Determination of perchlorate ions is possible in drinking/mineral water samples (after chloride precipitation); chloride precipitation does not affect perchlorate analysis. (4) There was good selectivity, long-term membrane stability, and applicability over a wide pH range. (5) Minimum detection limit was 0.8 ng/mL (ppb) for a 100 s irradiation time; the MDL is suitable for environmental analyses as the EPA DWEL is 24.5 ppb. (6) There is very good adhesion of the membrane and very low chemical reagents consumption. ’ AUTHOR INFORMATION Corresponding Author
*Phone: þ30 2821 037666. Fax: þ30 2821 037841. E-mail:
[email protected].
(9) Li, Z.; Xiao, L. F.; Deyhle, G. M.; Sesser, D. E.; Skeels, M. R.; Lamm, S. H. J. Occup. Environ. Med. 2000, 42, 200–205. (10) Clark, J. J. J. . In Perchlorate in the Environment; Urbansky, E. T., Ed.; Kluwer Academic/Plenum: New York, 2000; Chapter 3. (11) National Academy of Sciences. Health Implications of perchlorate ingestion, National Research Council of the National Academics, Prepublication Copy, 2005. (12) USEPA. U.S. EPA Headquarters Press Releases: EPA sets reference dose for perchlorate (retrieved March 14, 2005). (13) Weiss, J. A.; Stanbury, J. B. Anal. Chem. 1972, 44, 619–620. (14) Gallego, M.; Valcarcel, M. Anal. Chim. Acta 1998, 98, 1593–1687. (15) Batjoens, P.; De Brabander, H. F.; T’Kindt, L. Anal. Chim. Acta 1993, 275, 335–340. (16) Wagner, P. H.; Pepich, B. V.; Pohl, C.; Later, D.; Srinivasa, K.; Lin, R.; De Borba, B.; Munch, D. J. J. Chromatogr., A 2007, 1155, 15–21. (17) Sanchez-Pedreno, C.; Ortuno, J. A.; Hernandez, J. Anal. Chim. Acta 2000, 415, 159. (18) Shamsipur, M.; Soleymapour, A.; Akhond, M.; Sharghi, H.; Hasaninejad, A. R. Sens. Actuators, B 2003, 89, 9. (19) Segui, M. J.; Lizondo-Sabater, J.; Martinez-Mannez, R.; Sancenon, F.; Soto, J. Analyst 2002, 127, 387. (20) Huang, X.; Chai, Y.; Wand, X.; Li, Q. Anal. Sci. 2004, 20, 1185–1188. (21) Hatzistavos, S. V.; Koulouridakis, E. P.; Aretaki, I. I.; KallithrakasKontos, G. N. Anal. Chem. 2007, 79, 2827–2832. (22) Van Grieken, R. E.; Marckowicz, A. A. Handb. X-Ray Spectrom. 1993, 14, 463. (23) Vekemans, B.; Janssens, K.; Vincze, F.; Van Espen, P. Spectrochim. Acta, Part B 1995, 50, 149–169.
’ ACKNOWLEDGMENT Thanks are expressed to the Special Research Fund Account of the Technical University of Crete for the financial support of this research. ’ REFERENCES (1) Motzer, W. E. Environ. Forensics 2001, 2, 301. (2) Yu, L.; Canas, J. E.; Cobb, G. P.; Jackson, W. A.; Anderson, T. A. Ecotoxicol. Environ. Saf. 2004, 58, 44–49. (3) Smith, P. N.; Theodorakis, C. W.; Anderson, T. A.; Kendall, R. J. Ecotoxicology 2001, 10, 305–313. (4) Kirk, A. B.; Smith, E. E.; Tian, K.; Anderson, T. A.; Dasgupta, P. K. Environ. Sci. Technol. 2003, 37, 4979–4981. (5) Richardson, D. S. Anal. Chem. 2009, 81, 4645–4677. (6) Richardson, D. S. Anal. Chem. 2007, 79, 4295–4324. (7) Wolff, J. Pharmacol. Rev. 1998, 50, 89. (8) Soldin, O. P.; Bravermann, L. E.; Lamm, S. H. Ther. Drug Monit. 2001, 23, 316. 3391
dx.doi.org/10.1021/ac103295a |Anal. Chem. 2011, 83, 3386–3391
