294
Anal. Chem. 1981, 53, 294-298
data for Co(I1) in this sample (4) agree well with those obtained by lophine CL after Chelex treatment, although the uncertified NBS value for Co(I1) in this sample is 0.18 pg/g, a value 33% lower than that obtained by three independent methods with three independent operators. The ratios of Ca, Mg, and Fe to Co in the bovine liver are approximately 700, 3400, and 1500, respectively.
(4) Miller, R. J. Ingle, J. D., Jr., unpublished data, Oregon State University, Cowallis, OR, 1979. (5) Marino, D. F.; Ingle, J. D., Jr. Anal. Chem. 1979, 57, 2051-2053. (6) Kingston, H. M.; Barnes, I.L.; Brady, T. J.; Rains T. C.; Champ, M. A. Anal. Chem. 1978, 50, 2064-2070. (7) Hoyt, S.; Ingle J. D., Jr. Anal. Chlm. Acta 1976, 87, 163-175. (8) Montano, L. A.; Ingle, J. D., Jr. Anal. Chem. 1979, 57, 919-925. (9) "Analytical Methods for Flame Spectroscopy"; Varian Techtron: Palo Alto, CA, 1972. (IO) Leyden, D. E.; Patterson, T. A.; Alberts, J. J. Anal. Chem. 1975, 4 7 , 733-735.
LITERATURE CITED (1) Seitz, W. R.; Hercules, D. M. "Chemiluminescence and Bioluminescence"; Cormier, M. J., Hercules, D. M., Lee, J.. Eds.; Plenum: New York, 1973; pp 427-449. (2) Steig, S.; Nieman, T. A. Anal. Chem. 1977, 49, 1322-1325. (3) Montano, L. A.; Ingle, J. D., Jr. Anal. Chem. 1979, 57, 926-930.
RECEIVED for review July 4,1980. Accepted November 21, 1980. Acknowledgment is made to the National Science Foundation (Grant No. CHE 7617611 and 7921293) for partial support of this research.
Determination of Chromium(V1) in Water by Lophine Chemiluminescence D. F. Marho' and J. D. Ingle, Jr." Department of Chemistry, Oregon State University, Corvallis, Oregon 9733 1
The analytical utlllty of the lophlne-Cr(V1) chernilumlnescence (CL) reaction for the determlnatlon of Cr(V1) In natural waters is demonstrated. Optimization and callbratlon curve data are presented along wlth Interference data for 37 specles commonly found in natural waters. The use of an Ion exchange resin for CL sample preparatlon Is presented which allows the isolatlon or preconcentration of Cr( IV) from its potentlal lnterferents in less than 15 min. The llmlt of detection of Cr(V1) by this CL method Is 0.3 pg/L wlthout preconcentratlon and 0.015 pg/L wlth a 20-fold preconcentratlon. The llnear dynamlc range extends up to 100 mg/L, over 5 orders of magnitude.
The determination of Cr has received much attention (1-2.2). Table I is a summary of the pertinent results of the current analytical techniques being employed for the determination of Cr at ultra trace levels. Note that most methods (1) require a preconcentration step when applied to natural waters or (2) respond to both Cr(II1) and Cr(V1) or only Cr(II1). The first point is important because many of the preconcentration procedures are time consuming (a few hours to a day) and suffer from incomplete recovery such that standards must be carried through the preconcentration procedure. From the standpoint of environmental analysis, this last point is the most important since the speciation of Cr is critical and the specie most often of interest is Cr(VI), which has been associated with carcinogenic hazards (22,23) and is currently limited by the World Health Organization and the Safe Drinking Water Act to an absolute maximum concentration of 50 yglL in surface water intended for the abstraction of drinking water (24,25). Cr(II1) on the other hand is of substantially less concern and is in fact necessary for the maintenance of a normal glucose tolerance factor (26, 27). Since previous work in our laboratory with the lophine 'Present address: E. I. du Pont de Nemours and Co., Wilmington,
DE.
0003-2700/81/0353-0294$01 .OO/O
chemiluminescence (CL) system for Co had indicated a selective and sensitive response for Cr(V1) over Cr(II1) (28), it was decided to reoptimize the lophine system for maximum Cr(V1) response. It was felt that anion exchange could be used to separate Cr(V1) from most potential metal interferents which usually exist as cations in solution unless associated with ligands to form anionic complexes. Hence, a concurrent investigation was undertaken to determine the utility of a fast (e15 min) anion exchange sample cleanup procedure for the Cr(V1)-lophine system. Because of the difficulties of tailoring an analytical method to the specific determination of Cr(VI), several workers (5,15,18,19) have employed a strong anion (quaternary ammonium) exchange separation technique with conventional methods of Cr analysis for the purpose of speciation specificity. They have experienced problems with the quantitative recovery of Cr(V1) due to its tendency to remain on the resin unless such harsh treatment as conversion to Cr(II1) with Fe(I1) or elution with 4 N H2S04is employed. For this reason, it was decided to employ a medium strength (mixed quarternary-tertiary ammonium) anion exchange resin, Bio-Rex 5. Consisting of mixed R-N'(CH3)ZCl- and R-Nf(CH3)z(CzH40H)C1-groups, this resin is intended for use in the purification of organic acids (29). Its use as an inorganic ion exchange resin results in the quantitative recovery of Cr(V1) with much milder eluents than those employed with the strong base resin.
EXPERIMENTAL SECTION All CL measurements were obtained with the discrete sampling CL photometer system reported earlier (14), and with the modification and approximate instrumental conditions previously described (30). All solutions were prepared, unless stated otherwise, with Millipore (Milli-Q)reagent grade deionized water (MW) with reagent grade chemicals. Lophine (2,4,5-triphenylimidazole, Aldrich) was used without further purification. Solutions were prepared by dissolving a quantity of lophine in 25 mL of 2.0 M "OB and diluting to 250 mL with reagent grade methanol to yield a final HN03 concentration of 0.2 M. 0 1981 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 53, NO. 2, FEBRUARY 1981
295
Table I. Current Methods for the Determination of Cr method graphite furnace AA graphite furnace AAa flame AA flame A A ~ flame AAC flame A A ~ visible absorptiond (diphenylcarbizide) gas chromatographye gas chromatography liquid chromatography f X-ray fluorescenceg neutron activation differential pulse polarography luminol CL
preconcentration none Fe( OH), co-ppt 5-fold evaporn AF'DC-MIBK ext anion exchange MIBK ext MIBK ext HTFA-benzene ext HTFA-benzene ext none ion exchange none none
% recovery
92
96 no data 90 100 97
response Cr(II1) + CrlVI) . , cr(II1j Cr(II1) + Cr(V1) Cr(V1) Cr(VIj Cr(VI) Cr(V1) Cr(II1) t Cr(V1) Cr( 111) Cr( VI) Cr(111) Cr(II1) + Cr(V1) Cr( VI)
95 123
100
Cr(111)
none
detection limit, pg/L
ref
0.1 0.02 4.3 0.05 0.1 10 1.2
2 1 4 3
0.1 0.25 0.8
8
5 6 7
0.12 10
9 10 11 12 13
0.02
14
0.25
a Total Cr determined with addition of Fe(II), Cr(V1) found by difference, 1 day precipitation, 140 mL sample size. 0.08 L Sam le extracted into 20 m L of MIBK, 30 min oxidation to convert Cr(II1) to Cr(V1). 1 L sample, concentrated to 10 mL. hJCr(III) oxidized t o Cr(V1) for total Cr, applied to clinical samples. e Cr(V1) only partially extracted so reduced t o Cr(II1). f With coulometric detection, 20 min analysis time. g Cr(V1) reduced to Cr(II1) for total Cr, Cr(V1) found by difference, 30 min batch ion exchange.
Standard Cr(V1) solutions were prepared via quantitative dilution in MW from a 1001 mg/L Cr(VI) stock solution prepared by dissolving 2.8308 g of previously dried KZCrzO7(primary standard, Baker) in 1.0 L of MW. HzOzsolutions were prepared by dilution of 30% HzOz(reagent, Mallenckrodt, unstabilized) in M disodium dihydrogen ethylenediaminetetraacetic acid (EDTA) (primary standard, Fredrick Smith Co.). The addition of EDTA aided in low&g the blank CL signal by a factor of 5, presumably by complexation of the metal impurities contained in the HzOZ(31). No pH adjustment to the HzOzsolutions was necessary, as the optimum pH for stability of 4.0 was roughly achieved by the buffering action of the added EDTA. Bio-Rex 5 anion exchange resin (50-100 mesh, C1 form, Bio Rad Laboratories) was cleaned by washing three times in MW. Next, the resin was placed in a polypropylene ion exchange column (0.7 cm i.d. X 4 cm length, Bio Rad No. 731-1110), washed again with MW, and used without further purification. All other solutions (e.g., interferent cations, anions) were prepared as previously described (30). The general CL measurement procedure consisted of addition with Eppendorf pipets of the following quantities of the temperature equilibrated solutions (25 "C) to the reaction cell: 1.0-mL sample or blank, 0.5 mL HzOzsolution, and 0.5 mL lophine solution, followed by final injection of 0.5 mL of KOH solution with an automatic dispensing syringe to initiate the CL reaction. The CL signal was taken as the difference in the mean peak height between five sample and blank runs Cell cleaning, as before (28), proved critical to good reproducibility. A wash solution consisting of 1 M HN03 in methanol was employed throughout this study. A single rinse with this solution, followed by two MW rinses eliminated any memory effects and provided a relative standard deviation (RSD) of 2% at the 10 pg/L Cr(V1) level. The analyte and interferent detection limits ( q )are defined as the concentration of the analyte or interferent that produces a CL signal equal to twice the standard deviation in the blank signal, while the interference level (q*)is taken as the concentration of interferent required to produce a change in the Cr(V1) signal at the Cr(V1) detection limit equal to twice the blank standard deviation (evaluated at 3 pg/L Cr(V1)). The standard deviation of the blank signal was determined by the noise in the dark current and not by the irreproducibility of the blank signal. Graphite furnace atomic absorption (GFAA)data were obtained on a Perkin-Elmer Model 403 atomic absorption spectrometer with an HVA 2100 graphite furnace at a wavelength of 3578.7 A and cycle parameters of: 40 s dry, 120 "C; 40 s ash, 400 "C; 7 s atomize, 2500 "C. The Nz flow rate was 3.0 L/min; the gas interrupt was consistently off, and the sample size was 10 pL. Both CL, and GFAA measurements were performed on tap water and Willamette River water samples collected in 1-L
r 25
5.0,
t-
k. 2:
10
I \/
I
016
@-]
I
O!8
l!O
M
Figure 1. KOH optimization: [Cr(VI)] = 30 pg/L, [lophine] = 4 X M, [H202] = 0.1 M.
Nalgene CPE poly bottles which had previously been cleaned with 50% HN03and rinsed 15 times with MW. In addition, the river water sample was filtered through a 1.2-pm Millipore filter t o remove particulate matter prior to any analysis. CL and GFAA measurements were made within 48 h of each other to minimize the chances of Cr(V1) adsorption on the container walls. All reagent concentrations are reported as final cell concentrations after dilution, unless otherwise stated, whereas analyte and interferent concentrations are reported as initial concentrations. All CL signals are given in terms of photoanodic current.
RESULTS AND DISCUSSION Optimization Studies. Figure 1illustrates the results of the KOH concentration optimization study. As can be seen, no real peak or plateau in CL response is evident. Examination of the blank signal, however, reveals a region of approximately constant background signal between 0.1 and 0.8 M KOH concentrations (cell concentration), followed by a rise in the blank signal at higher KOH concentrations. For this reason, as well as the inconvenience of preparing highly concentrated KOH solutions, an optimum point was chosen at 0.8 M KOH (this represents a precell concentration of 4.0
MI.
296
ANALYTICAL CHEMISTRY, VOL. 53, NO. 2, FEBRUARY 1981
5c)
1
,SO
.40
4 30
e 4 213
20
- R1 ANY BLANK 113
10
-+----*-DL
0 -3.0
-210
Figure 2. H202optimization: M, [KOH] = 0.8 M.
OJ
-4!5
[Cr(VI)]
= 30 pglL,
3!5
V!O LOG
-1!o
-115
3!0
O6
0
[lophine] = 4 X
Io
-l!5
EOPHINE~ ,M
Flgure 3. Lophine optimization: M, [H202] = 0.04 M.
[Cr(VI)]
= 30 pg/L,
[KOH]
= 0.8
Figure 2 shows the results of the H z 0 2concentration optimization study. In this case, a definite plateau in the Cr(V1) CL signal exists a t a HzOz cell concentration of 0.02 M or greater. To ensure a constant CL signal level in the event of some small amount of HzOz decomp,,ition with time, an optimum HzOzcell concentration of 0.04 M was chosen (0.2 M precell concentration). Figure 3 illustrates the results of the lophine concentration optimization study. Again, a plateau in CL response is evident a t lophine concentrations greater than ca. 2 X M (cell concentration). To conserve lophine and still operate on this plateau, we chose a final optimum cell concentration of 4 X M (2 X M precell concentration). The acidity of the lophine solution deserves some mention at this point. When acidified lophine is added to a Cr(V1) sample in the presence of HzOz,the peroxo species CrO(02)2 is formed (32) HCr04- + 2 H 2 0 2 + H+
-
Cr0(0J2
+ 3Hz0
(1)
as evidenced by the formation of an intensely blue species in the reaction cell when Cr(V1) concentrations greater than 50 mg/L are used. Note that an acidic environment is required for this reaction to proceed. A pH study of this pre-CL reaction indicated that a maximum CL signal was achieved
/ 1’0
2’0
3!0
4’0
510
610
ANALYTICAL CHEMISTRY, VOL.
53, NO. 2, FEBRUARY 1981
297
Table 11. Potential Interferents in the Cr(V1)-Lophine CL Systema species
c1, mg/L
m, nA mg-' L-I
C1*
M , C nA mg-' L-l
Mg( 11) Ca(11) Fe( 111) Fe( 11) Cr( 111) Co(11) Se(IV) V(V) Ga(111) CU( 11) Al( 111) c1BrIMn0,oc1Fe( CN)63Au( C1),humic acid
150 8.5 2.4 2.7 0.05 0.001 300 0.86 15 24 3.7 >10000
-6.7 X 1.4 X lo-' 5.0 X lo-' 4.5 x 1.7 100 2.7 x 10-4 5.6 X lo-' 5.4 x 10-3 1.6 x 10-3 2.1 x
55 20 3.2
-1.8 x 10-3 5.9 x 10-3 4.0 X lo-'
0.03 0.035 0.02 2.5 8.7
2.7 2.3 4.7 x 10-3 4.9 x lo-' 9.2 x 10-3
>loo >loo
0.017 0.004
1000 8.6 9.6 0.01 0.05 9.3
4.7 32
-8.4 x 10-5 -1.2 x lo-' -1.0 x 7.8 1.6 -8.6
X
Si03'-, Se04'-, As(V), As(III), Zn(II), a Species for which c1> 100 mg/L: Sn(VI), Na(I), K(I), Ti(IV), Cd(II), Sr(II), Rb(1). Slope of calibration curve near the detection limit, for Cr(V1) m equals 1.3 x l o 2 nA mg-' L-I and Slope of enhancement or depressive curve, ACr(V1) CL signal/A [interferent] with 3 pg/L Cr(V1). c1= 0.3 ppb. determination in previous studies (28,331and Cr(V1) in this study. Mutual interference is not a problem because reagent conditions are adjusted differently in both procedures for the best detection of the appropriate analyte. A cation exchange cleanup step is essentially mandatory for the lophine CL Co(I1) determination (33) because of several interferences and elimates the Cr(V1) interference if the Cr(V1) concentration is above its 5-pg/L interference level. Although the interference level for Co(I1) is quite low (4 pg/L) under the Cr(V1) determination conditions, it is still above the typical levels of Co(I1) in water samples and Co(I1) can be eliminated by the anion exchange procedure to be discussed if its concentration is abnormally high. Of the anions listed, only OC1- (present in tap water a t typical levels of ca. 0.4 mg/L) appears to slightly interfere. Fortunately, however, OC1- decomposes rapidly in an open container or is not retained on the anion exchange resin employed in this work, as will be demonstrated later. Hence, when coupled with an anion exchange cleanup routine, there are no apparent interferences in this method. For Br- and I-, c1* < c1 indicating these two halide ions do not affect the blank but in the presence of Cr(V1) decrease the Cr(V1) signal possibly by reducing some of the Cr(V1). However, such reactions would have already occurred in real samples. Ion Exchange Studies. Although very little data are available from Bio Rad concerning Bio-Rex 5, the similarity of this resin to strong (quarternary 3mmonium) anion exchange resins allowed some initial assumptions to be made as to a selectivity order. For strong resins (e.g., AG 1-X8, Dowex 1-X8, IRA 400, etc.) the selectivity order for some singly charged anions is F- < OH- < Ac- < IO3- < HC03- < C1- < NO2- < HS03- < CN- < Br- < NO3- < C103- < HS04< phenylate < I- (29). This order should be roughly followed by a mixed tartiary-quaternary ammonium group resin. Hence, to ensure quantitative removal of the Cr(V1) present in the samples, the resin was converted to the easily displaced OH- form. Artificial samples containing 100 pg/L Cr(V1) were placed on the resin, washed with 20 bed volumes of MW (to simulate real sample washing), and eluted with different displacing solutions. The eluents were subsequently analyzed via lophine CL to determine the Cr(V1) yield. Of the strongly retained anions listed above, eluents of potential buffers or reducing agents were deemed unacceptable and were not investigated, due to their tendency to introduce subsequent pH interference
in the CL analysis or reduce Cr(VI), respectively. These restrictions narrowed the field of potential eluents considerably. The final procedure utilized was as follows. One milliliter of resin was placed in the column. This was converted to the OH- form with 10 mL of 1M NH40H and washed with 10.0 mL of MW to lower the pH of the resin to