2090
Anal. Chem. 1982, 5 4 , 2090-2094
Interferences in Automated Phenol Red Method for Determination of Bromide in Water Chrls L. Basel and James D. Defreese" Department of Chemjstry, The Unlversity of Kansas, Lawrence, Kansas 66045
Donald 0. Whittemore" Kansas Geological Survey, The University of Kansas, Lawrence, Kansas 66045
The phenol red method for the determlnatlon of bromide In water has been automated by segmdnted flow analysls. Samples can be analyzed at a rate of 20 samples/h wlth a method detection Ilmlt, deflned as the cdncentratlon glvlng a signal about three times the standard deviation of replicate analyte determlnatlbns In reagent water, of 10 pg/L. Samples studled Include oil-field brines, hallte solutlon brlnes, groundwaters contaminated wlth these brines, and fresh groundwaters. Chlorlde and bicarbonate cause slgnlflcant positive interferences at levels as low as 100 mg/L and 50 mg/L, respectlvely. Ammonia gives a negatlve Interference that is Important at levels as low as 0.05 mg/L. An ionic strength buffer Is used to suppress a posltlve lonlc strength Interterence, correction curves are used to compensate for the chloride Interference, the blcarbonate Interference Is mlnimlzed by acldlficatlon, and the ammonia Interference is eliminated by Its removal by Ion exchange. Reactlon product studles are used to suggest a plauslble mode of chlorlde interference.
and concentration. Chloride was of major concern here because it may be found in high concentrations in oil-field brines, halite solution brines, and freshwaters polluted with either of these. Ammonia, which is often found in oil-field brines and freshwaters, has been mentioned as an interference in the phenol red method (9,10,12),but no quantitative data have been published. Bicarbonate is often a major constituent in groundwaters. Reports of its tendency to interfere have been conflicting (9, 11, 13). Other potential interferents are iodide (7,9,12), nitrite (12, 13),ferric iron (13),and manganese (13). These either exhibit negligible effects (5,13) or are normally found at such low concentrations in natural waters that they are not important to our studies. In this paper, the automation of the phenol red method for the determination of bromide in natural waters is described. Chloride, ammonia, and bicarbonate interferences are quantified and compensation procedures are presented. Studies of the chemistry of the phenol red method to indicate the actual products of the reaction and the mode of chloride interference are also discussed.
The concentration ratio of bromide to chloride has been established as indicative of the source of sodium chloride contamination of natural waters (1,2). At a given chloride concentration, this ratio is larger for oil-field brines or waters polluted by oil-field brines than it is for halite solution brines or waters contaminated by halite solution brines. A rapid, accurate method for the determination of bromide was needed for further study and application of this phenomenon. Colorimetric methods were examined because they are well-suited for automation by continuous flow techniques such as segmented flow analysis (SFA) or flow injection analysis (FIA). An SFA automated method based on the catalytic effect of bromide on the oxidation of iodine to iodate by potassium permanganate in sulfuric acid solution (3) exhibits an unacceptable sensitivity to chloride ( 4 , 5 )which precludes its use for the saline samples of interest here. The standard phenol red method (6) indicated sufficient accuracy, precision, and sensitivity for our studies. This method has also been shown to be amenable to automation by SFA (7). Preliminary results of the automation of the phenol red method via FIA are encouraging but are not discussed in this paper. A fluorescein method for bromide that may have a low sensitivity to chloride and a low detection limit is presently being investigated by others (8). The latest edition of ref 6 mentions that interferences to the phenol red method may be present in saline or polluted waters (6),but it does not identify them. Various levels of chloride have repeatedly been stated not to interfere in the phenol red method (7, 9, lo), although one report (11) has warned of a possible interference depending on reaction time
EXPERIMENTAL SECTION Apparatus. A Technicon AutoAnalyzer I1 consisting of an autosampler, proportioning pump, appropriate analytical cartridge, spectrophotometer, and strip chart recorder was utilized for the automation of the phenol red method by SFA, the interference studies, and the ammonia and chloride determinations. A Perkin-Elmer 555 UV-visible spectrophotometer was used t o acquire spectra of reactants and products of the reaction. Reagents. All reagents were analytical reagent grade. Water for solution preparation and sample dilution was deionized and distilled. M The phenol red and chloramine-?' solutions were 3.9 X and 4.6 X lo-, M, respectively. The ionic strength/pH 4.6 buffer contained 2.1 M acetic acid, 2.0 M sodium acetate, 1.5 M sodium nitrate, 0.4 M magnesium sulfate, and 2 mL/L Brij-35 surfactant. A 6 M sodium nitrate solution was used t o recharge the ion exchange column. A "dilute" pH buffer that provides a relative reagent concentration in the flow stream similar to that in the manual method (6) was used for several studies. It contained 0.35 M acetic acid, 0.34 M sodium acetate, and 2 mL/L Brij-35 surfactant. For the spectral studies, the buffer, chloramine-T, and phenol red solutions were prepared as for the manual method (6). A 6.1 X 10" M bromophenol blue solution was also used. For the interference studies, the following reagents were used to prepare stock solutions: NH4C1for NH,, Ultrex NaCl for C1, NaHCO, for HC03, NaN03 for Na and NO3, Na2S04for SO4, MgSO, for Mg, and Ca(N03),.4H20for Ca. Bromide standards were prepared from KBr. Automated System. A schematic diagram of the SFA system for bromide determination is shown in Figure 1. Pump tube diameters (flow rates) and reagent concentrations were chosen such that the reagent concentrations in the flow stream were
0003-2700/82/0354-2090$01 2510
0 1982 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 54, NO. 12, OCTOBER 1982
2091
Table I. Effect of Major Constituents of Natural Waters on the Determination of Bromide (0.50 mg/L) by the Phenol Red Method
constituent FROM DEBUBBLER
Na NO3 Mg
so 4 Ca
c1 Figure 1. Schematic diagram of segmented flow analysis system for determination of bromide.
HCO,
3 "
concn, mg/L 700 500 3 50 1500 700 1000 1000 0.1
measd [bromide], mg/L dilute ionic strength/ pH buffera pH buffera 0.61 0.53 0.60 0.65 0.62 0.87 >1.50b 0.37c
. I
0.49 0.49 0.49 0.48 0.49
0.58 1.08 0.43
See text for composition, Off scale. These values were determined with the ion exchange column removed. a
similar to those in the manual method reaction mixture (6). However, the buffer was increased in capacity and ionic strength to minimize interferences due to variable buffer capacities and ionic strengths of the mmples. A 20-per-hour adjustable cam was modified to increase the wash cycle time by cutting off one section of both sample positions. This reduced carryover. The sample is first pumped into a debubbler to remove the bubble introduced whlen the sample probe changes from sample to wash and vice versa. This debubbling process eliminates base line oscillation, a problem also noted by others ( 4 ) . Next, the sample enters an ion exchange column (Dowex 50WX8) to remove ammonium ion. After this column, the segmenting air bubbles are introduced. Next, buffer, phenol red, and chloramine-T solutions are introduced in sequence and mixed by coils. After the chloramine-T introduction, additional coils delay the absorbance measurement by 8 min, a time chosen because the maximum net absorb,snce above the base line is nearly reached at this time. At longer times, the base line absorbance becomes too large to be zeroed on the recorder. The solution is then debubbled and enters the spectrophotometer flow cell where the absorbance is measured at 590 nm with an 8-nm band-pass. General Procedure. Chloride and ammonia were determined by Technicon methods which use a modified Volhard procedure (14) and the Berthelot reaction (15),respectively. For the bromide determination, the recorder is zeroed with the sampler in the wash cycle. The highest standard is then sampled several times to set full scale. A series of standards is placed first on the sample carousel, followed by samples with water interspersed every seven to eight samples. Approximately every hour, the ion exchange column is recharged by sampling two aliquots of the 6 M sodium nitrate solution. This highly concentrated solution causes extreme base line noise and depression due to ammonia release. Therefore, two aliquots of water are sampled immediately before anid after the sodium nitrate. Standards are run at approximately 1.5-h intervals.
RESULTS A N D DISCUSSION System Characterization. Bromide standards were run in both ascending and descending order with respect to concentration. The calibration curves are alpproximatelycollinear with an average equation of peak heiglht (arb units) = (60.70 f 0.34) [Br-] + (0.06 f 0.25), with an SEE (standard error of estimate) = 0.46, and r (correlation coefficient) = 0.99991, over a range of 0.026-1.5 mg/L bromide. The detection limit, defined as that concentration giving a signal equal to the Students t value for a one-tailed test a t the 99% confidence ltsvel with N - 1degrees of freedom times the standard deviation of repetitive determinations of bromide in a 25 gg/L standard (16), is 10 gg/L. This is also the concentration giving signal greater than the blank measure by three times the standard deviation of the blank determinations ( 1 7 ) . Therefore, the phenol red method is suitable for the determination of bromide in most groundwater and surface water. Precision was meamred by determining the bromide concentration of each of 26 samples nine times during a 4-day period. These samples consisted of fresh and saline ground-
waters and brines with bromide concentrations ranging from 0.14 to 1.1mg/L, after any necessary dilution. Percent relative standard deviations ( % RSD) ranged from approximately 1% at the higher concentration to 8% at the lower concentration. An indication of the accuracy of the method was obtained by spiking samples with 0.25 mg/L Br-. The samples consisted of fresh and saline groundwaters and brines, with bromide concentrations ranging from approximately 0.15 to 1.0 mg/L. The average recovery was 98.4 rfr. 4.9% with a range of 91.2 to 106%. Identification of Interferents. For identification of potentially serious interferences, bromide concentrations were repeatedly determined in solutions that were spiked with 0.5 mg/L bromide and that contained the common major cations and anions found in freshwater and/or saline water. The major constituent concentrations used are near the maximum found in freshwaters or saline waters diluted to bring the bromide concentration within the range of the method and/or to reduce the chloride interference. The results are given in Table I. Use of the dilute pH buffer gave a positive interference from all major constituents added except ammonia, which, as expected, gave a serious negative interference. The positive interferences were roughly linear with increasing ionic strength of the added constituents. By use of the manual method (6), these interferences were found to be important a t 8 min but not a t 20 min for all positive interferents except chloride and bicarbonate. The chloride interference was greater at 20 min than at 8 min. Use of the ionic strength/pH buffer decreased the positive interferences to insignificant levels except for chloride and bicarbonate, both of which still caused substantial positive interferences. Therefore, chloride, ammonia, and bicarbonate were selected for further, more detailed studies. Chloride Interference. The extent of the chloride interference is shown in Figure 2. A significant positive interference is seen for chloride levels as low as 100 mg/L. The curves in Figure 2 were obtained by adding known amounts of chloride to bromide standards and measuring the bromide concentration, which is the sum of the actual bromide and the apparent bromide due to chloride. The actual bromide concentration in the solution is then subtracted from the measured bromide concentration to give the error (correction) due to chloride. These curves can be used to correct for the chloride interference. The chloride and bromide in the sample are determined in separate measurements. The appropriate correction is then related to the measured bromide by the correction curve (or interpolation between curves) corresponding to the chloride concentration in the sample. The correction is then subtracted from the measured bromide concentration
2092
ANALYTICAL CHEMISTRY, VOL. 54, NO. 12, OCTOBER 1982
0
,
i---1 -
1
Table 111. Effect of Ion Exchange on the Determination of Bromide and Ammonia in Bromide Standards Spiked with 0.5 mg/L Ammonia and in Natural Waters
'L
m
[bromide], mg/L in standard 0.05
L
2L
W
measd concns, mg/L without ion exchange with ion exchange ammonia bromide ammonia bromide
0.10
o
?
0
.25
1
.75
5
Measured Br-,
7.2,'
1.5
0.25 0.50 1.00 1.50
mg/L
Flgure 2. Error in the determination of bromide caused by chloride: (A) 2000 mg/L CI-; (B) 1000 mg/L CI-; (C) 500 mg/L CI-; (D) 200 mg/L CI-; (E) 100 mg/L CI-.
Table 11. Effect of the Chloride Correction on the Determination of Bromide in 0.50 mg/L Bromide Standards Spiked with Chloride and in Natural Waters mg'L in bromide standards
measd [bromide], mg/L uncorrected correctedu
50
0.50 0.52 0.55 0.60 0.64
100
500 1000
2000
sampleb OFBl F 1 + OFB2 HSBl F 2 t HSB2 F3
0.50 0.50 0.50 0.50 0.50
measd [bromide], mg/L [chloride], uncor- cordilutionC mg/Ld rected rected' 1/100 1/100 1/100 10/100 none
285 277 1930 948 25
0.50 0.50 0.49 0.50 0.50 0.50
1.13 1.00 0.34 0.24 0.14
1.07 0.94 0.23 0.18
0.14
Correction derived from data shown in Figure 2. OFB = oil-field brine; F = fresh groundwater; HSB = Dilution of sample prior to brohalite solution brine. Determined separately by method mide determination, given in ref 14. a
to give the actual bromide concentration. The effect of the correction can be seen in Table 11. The correction works quite well for standards. For natural water samples the correction is good to within a few percent or better as indicated by the recovery study. A much higher chloride interference was seen when the dilute pH buffer was used. This indicates that the chloride interference consists of both an ionic strength component (eliminated with ionic strength/pH buffer) and an effect intrinsic to chloride. A plot of apparent bromide due to chloride vs. chloride for chloride standards should not be used as a correction curve, as it has been for the catalytic oxidation method (4),because the correction depends on both the chloride and actual bromide concentrations. This may be more easily seen by replotting the data of Figure 2 as measured bromide vs. chloride. Curves drawn through the points with the same actual bromide exhibit increasing slopes for each increment in actual bromide. To compensate for a chloride interference in seawater, a set chloride level has been added to bromide standards (11). However, this would not give an accurate measurement here
sample6
dilutionC
OFBl F1 + OFB2 HSB 1 F2 + HSB2 F3
1/100 1/100 1/100 10/100 none
