Fixed-site ion exchanger for liquid chromatographic determination of

Gore-Tex membrane; Harry C. Dorn of VPI & SU for the loan of one of the IBM LC pumps;and Don Stingof Spectra-Tech,. Inc., for use of theBarnes Model 6...
0 downloads 0 Views 781KB Size
Anal. Chem. 1985. 57, 615-620

extractor/separator portion of this apparatus. In spite of this, readily identifiable, high signal-to-noise, on-line spectra are obtained with less than 300 Hg of injected sample. Carbon tetrachloride has been demonstrated to possess all of the necessary characteristics in order for it to be employed as an infrared-transparent,flowing-extraction solvent. Thorough investigation of this RP-HPLC/FTIR method is currently under way.

ACKNOWLEDGMENT The authors wish to thank the following people for their assistance: Robert A. Sanders of Procter and Gamble for discussions at the inception of this RP-HPLC/FTIR approach; Gil Pacey of Miami University for encouraging discussions on flow injection analysis and for supplying a sample of the Gore-Tex membrane; Harry C. Dorn of VPI & SU for the loan of one of the IBM LC pumps; and Don Sting of Spectra-Tech, Inc., for use of the Barnes Model 600 beam condenser.

LITERATURE CITED (1) Johnson, C. C.; Taylor, L. T. Anal. Chem. 1983, 55, 436-441. (2) Brown, R. S.; Taylor, L. T. Anal. Chem. 1983, 55, 1492-1497. (3) Amateis, P. G.; Taylor, L. T. Anal. Chem. 1984, 56,966-971.

615

(4) Brown, R. S.; Amateis, P. G.; Taylor, L. T. Chromatographia 1984, 78, 396-400. (5) Johnson, C. C.; Taylor, L. T. Anal. Chem. 1984, 56, 2642-2647. (6) Vidrine, D. W. I n “Fourier Infrared Spectroscopy”; Ferraro, J. R., Basile, L. J., Eds.; Academic Press: New York, 1979; Vol. 2, pp 129-164. (7) Jinno, K.; Fujlmoto, C. Chromatographla 1983, 77, 259-262. (8) Parris, N. A. J . Chromatogr. Sci. 1979, 77, 541-545. (9) Jinno, K.; Fujimoto, C.; Uematsu, G. Am. Lab. (Fairfield, Conn.) 1983, 39-45. 10) Kuehl, D. T.; Grlffiths, P. R. J . Chromatogr. Sci. 1979, 77, 471-476. 11) Kuehl, D. T.; Grlffiths, P. R. Anal. Chem. 1980, 52, 1394-1399. 12) Kalasinsky, K. S.; McDonald. J. T., Jr.; Kalasinsky, V. F. FT-IR Spectral Lines 1983, 5 , 14-15. 13) Duff, P. J.; Conroy, C. M.; Griffiths, P. R.; Karger, B. L.; Vouros, P.; Kirby, D. P. Proc. SOC.Photo-Opt. Instrum. Eng. 1981, 289, 53-56. 14) Karlberg, B.; Thelander, S. Anal. Chim. Acta 1978, 9 8 , 1-7. 15) Conroy, C. M.; Grlffiths, P. R.; Duff, P. J.; Azarraga, L. V. Anal. Chem. 1984, 56, 2636-2642. 16) Nord, L.; Karlberg, B. Anal. Chim. Acta 1980, 778, 285-292.

LECEIVED for review October 1, 1984. Accepted December 13, 1984. The IBM LC/9533 chromatograph was a donation to VPI & SU from IBM Instruments, Inc., through their University Gifts program. The financial support of the Department of Energy through Grant No. DE-FG22-84PC70799and the Commonwealth of Virginia is appreciated.

Fixed-Site Ion Exchanger for Liquid Chromatographic Determination of Multifunctional Carboxylic Acids R. M. Cassidy* and S. Elchuk General Chemistry Branch, Atomic Energy of Canada Limited, Research Company, Chalk River Nuclear Laboratories, Chalk River, Ontario, Canada KOJ 1JO

Reversed phases coated with a “permanently” sorbed ion exchanger and indirect UV detection have been investigated for the determination of simple and muHHunctionai carboxylic acids in chemical cleaning solutions. The advantages of being able to vary both the ion-exchange capacity and the hydrophobic interactions on these types of ion exchangers for the optimization of resolution and detection are illustrated, and the selection of optimum separation conditions is discussed. Dissolved iron interferes with the analysis due to photochemical, redox, and kinetic effects but good recoveries can be obtained after reduction of the iron with hydroxylamine and complexation with 1,2-diaminocyciohexanetetraaceticacid. Detection limits (3X base line noise) for oxalate, citrate, ethylenediaminetetraacetate,and hydroxyethyienediaminetriacetate are 0.6-20 pgmL-’ for a 20-pL sample, and relative standard deviations are 3 to 5 % in the 75-350 pgmL-’ range. Analysis results for reactor decontamination solutions containing up to 250 pgmL-’ of iron agree with results obtained by other techniques, and it is shown that this technique should also be useful for determination of metal ions in the samples. A determination of the above reagents In the presence of Fe(I1) and NI(I1) takes 7 to 12 mln after a 5 to 10 min reduction step. Cr( I I I)forms nonlablie complexes with ethyienediaminetetraacetlc acid, and Its presence will cause low results for this acid.

Multifunctional carboxylic acids are used in formulations for boiler treatment, chemical cleaning of process systems, and

decontamination of nuclear reactor components. Our main interest in the determination of these reagents has been related to their use for reactor decontamination. Previously we developed a gas chromatographic procedure (1) based on the separation of methylated esters of the acids (2). However, this method requires a derivatization step, which makes the method relatively slow and susceptible to errors, and the use of gas cylinders can make this method cumbersome for in-field analysis. High-performance liquid chromatography (HPLC) has a number of potential advantages for these analyses, but the techniques reported thus far (3-8) are suitable for only a limited number of acids. Most of the methods (4-8) are based on the separation of metal chelates of the acids and are only applicable to acids forming stable chelates. These techniques also limit the choice of eluent pH and buffers, due to possible precipitation of metal ion added to form the chelates. We had previously examined ion pairing and metal chelate (Fe and Cu) HPLC separations for the determination of these acids but had only limited success. Our recent studies with reversed-phase columns that are “permanently” coated with large hydrophobic quaternary amines (9) to give a charged surface that can be used for ion exchange, have shown that such columns can offer distinct advantages over either dynamic ion pairing or bonded-phase ion-exchange separations. This paper reports the results of studies of these columns with both direct and indirect UV detection for the determination of multifunctional carboxylic acids. The main acids of concern in this study were oxalic, citric, hydroxyethylethylenediaminetriacetic (HEDTA), and ethylenediaminetetraacetic (EDTA) acids. Decontamination solutions can also contain large concentrations of iron (up to 250 pg-

0003-2700/85/0357-0615$01.50/00 1985 American Chemical Society

616

ANALYTICAL CHEMISTRY, VOL. 57, NO. 3, MARCH 1985

Table I. Eluents Used for Indirect Detection

compound

source"

4-hydroxyphthalic acid

A

benzenesulfinic acid (Na salt) 1,2,4,5-benzenetetracarboxylic acid

B C

1,2,3-benzenetricarboxylicacid 1,2,4-benzenetricarboxylicacid

C C

1,3,5-benzenetricarboxylicacid benzoic acid phthalic acid (mono K+ salt) salicylic acid

C D E E

rn-sulfobenzoic acid (mono Na+ salt) o-sulfobenzoic acid (mono NH, salt) 5-sulfosalicylicacid

F F G

wavelength maxima, nm

molar absorptivity, L-mol-'.cm-'

molar absorptivity at 254 nm, L.mol-'-cm-'

280 245 260 295 245 280 285 240 280 265 280 295 230 273 265 295

1300 7100 1200 2000 9000 750 1600 9500 1000 560 1400 3600 7200 1100 1300 3000

5800 960 7500 1200 6300 2500 660 1300 300 900 880 440

"A Alfa Products; B, Eastman Kodak; C, Aldrich Chemical Co., D, ICN Pharmaceuticals; E, Fisher Scientific; F, Pfaltz & Bauer hc.; G, J. T. Baker Chemical Co. mL-') as well as smaller concentrations of some other metal ions (chromium and nickel in particular), and t h e effects of these metal ions on the analyses have also been examined,

EXPERIMENTAL SECTION Apparatus. The chromatographic system consisted of M6000A Waters pumps, a Beckman 332 pumping system, a Rheodyne sampling valve, a Schoefell SF 770 variable wavelength detector, and a Spectra Physics 4270 computing integrator. The columns used were as follows: 10 pm PRP-1, 4.1 X 150 mm (Hamilton co.); 5 pm Supelcosil Lc-18, 4.6 X 150 mm (Supelco Inc.); and 3 pm Supelcosil LC-18, 4.6 X 75 mm. Reagents. The organic anions used both for elution and as chromophores for indirect detection of the multifunctional organic acids are listed in Table I; these reagents were used as received from the supplier (Table I). The tetrabutylammonium hydroxide (TBAOH) was purified by placing a 40% (w/v) aqueous solution in a freezer a t -20 "C until -80% of the volume was precipitated, and the precipitated TBAOH was washed with water at 0 "C and then analyzed by titration with 1 mo1.L-' HC1. The 1,2-diaminocyclohexanetetraacetic acid (DCTA) was purified by precipiatation of the DCTA from a basic solution by slow addition of HCl to give a final p H of 2. All other reagehts were prepared from HPLC-grade solvents and water purified by triple distillation or double distilled water purified in a Milli-Q (Millipore) system. Column Coating. The column coating procedure given here is for a 3 pm Cu reversed phase, the column recommended for this analysis. Coating procedures for the other columns are similar, and have been discussed elsewhere (9). A precolumn and analytical column were connected in series and washed with 1.4 L of 5 X lo4 mo1.L-l cetylpyridinium chloride in 23% (v/v) acetonitrile-water; the precolumn was a guard column that contained PRP-1 or Cl8 bonded phase, depending on which type of analytical column was used. The eluent was changed to that used for the analytical separation (contained 1 to 7% (v/v) acetonitrile and one of the acids in Table I; typical eluents are given in Tables I1 and 111) and the column was equilibrated with 150 mL of this eluent. The precolumn was placed upstream of the injector to ensure that cetylpyridinium was not eluted from the analytical column, ahd samples of EDTA were injected until a constant peak height was obtained. S t a n d a r d Preparation. T o simulate decontamination solutions, iron powder was added to a helium degassed solution containing EDTA, oxalic acid and citric acid, and stirred and heated under helium in an opaque container a t 50 "C until the iron powder was dissolved (-30 min). This solution was then stored in a sealed volumetric in the dark. A typical standard contained 300,400, 300, and 160 ppmL oxalate, EDTA, citrate and Fez+,respectively. Some standards were also made directly from the free acids with NaFe"'EDTA (Alfa Products) or FeC1, used as the source of iron. Reagent-grade chemicals were used

-

Table 11. Detection Limits

eluent sulfosalicylateb phthalate' 1,2,4-benzenetricarboxylatec 1,3,5-benzenetricarboxylate'

detection limit" wavelength, WrnL-' pg injected nm Ox CIT EDTA Ox CIT EDTA 295

0.8

4

3

0.02 0.08

0.06

254 257

2

0.6

9 5

5 3

0.04 0.2 0.01 0.1

0.1 0.06

254

2.1

20

5

0.04 0.4

0.1

OThree times peak-to-peak base line noise. *PRP-l column. c 5 bm CIRcolumn. Table 111. Values of k' for Different Eluent Systems anion diglycolate tartronic ketomalonic gluconic HEDTA maleic oxalate FeEDTA EDTA citrate DCTA

A

B

eIuent" and k' values C D E

F 3.6 4.0

4.3 1.9 2.0

21

3.5

30 15

14 23

3.0 9.3 12 27

2.8

1.9

3.9

2.0 1

16.8

7.5 8.8 15.4

22

37

6.9 4.4 6.3 9.1 11.5

mo1.L-' 1,3,5-tricarboxylic acid, pH 7.3, 80 mg "A, 1.0 X mol.L-' benzoic acid, pH 7.3, 80 coating on PRP-1; B, 3.0 X mg coating on PRP-1; C, 5.0 X lo-, mol.L-' 1,2,4-tricarboxylicacid, mo1.L-' phthalic . pH 7.3, 23 mg coating on PRP-1; D, 1.5 X acid, pH 7.3, 23 mg coating on PRP-1; E, 5 X lo-, mol.L-' sulfomol-L-' salicylate, p H 7.3, 23 mg coating on PRP-1; F, 5 X sulfosalicylate,pH 5.6, 80 mg coating on PRP-1. All eluents except F contained -1 X lomamol-L-' THAM buffer. for preparation of the standards. Sample Analysis. When possible, sample bottles were opaque and the sample was protected from exposure to light. One milliliter of 3000 pgrnL-l DCTA, which had been adjusted to pH 4 with tris(hydroxymethy1)aminomethane (THAM), 15 pL of 100 pg.mL-' NH20H.HC1, and 0.5 mL of the sample were added to

ANALYTICAL CHEMISTRY, VOL. 57, NO. 3, MARCH 1985

r

I

t

I

ic-----.*

I

1

I

75

s

I

I

OX

I

1

:I FEDCTA _L

I .1 11

I/

I

. -

RETENTION TIME

Figure 1. Separation of oxalate, citrate, and EDTA anlons In presence of iron and nickel. Experimental conditions: 4 X 75 mm, 3 pm CIB column coated with 5 X mo1.L-I cetylpyridinium chloride In 23% (v/v) acetonitrile; eluent, 1.5 X moi-L-‘ phthalate, 1.8 X lo-$ m0l.L-l THAM buffer, 7 % (v/v) acetonitrile, and flow rate of 2 mL.rnin-’; sample, 20 pL of solutlon containing 100, 117,67,33, and 100 pg-ml-’ of oxalate (OX), EDTA, citrate (CIT), NI(II), and Fe(III), respectively; detection at 254 nm.

a 4-mL scintillation vial equipped with a Teflon-lined cap. The vial was capped and heated for 5-10 min at 100 “C in a water bath that was covered with aluminum foil. The sample was cooled for a few seconds under a stream of water, and -60 pL of 1molC’ NaOH was added to give a pH of 7.3-8.0. An aliquot (normally 20 rL)was then injected into the chromatograph; chromatographic conditions recommended for most samples are those given in Figure 1. Quantitation was done by comparison of peak areas with those for standards reacted similarly.

RESULTS AND DISCUSSION Detection. Because of the variation in the physical and chemical properties of the different acids used for chemical cleaning, it is difficult to find one property that can be used to monitor the elution of all the reagenta, particularly at low microgram per milliliter concentrations. The methods examined in this work were detection in the low UV range (215 nm), and indirect (vacancy) detection. Many of the acids studied absorbed light in the low UV region, but there were wide variations in molar absorbtivities. Detector sensitivity was especially poor for citrate (-400 pgrnL-’ for 3 x base line noise) and sensitivity improvements obtained at lower wavelengths could not be used due to the presence of large background absorption from small concentrations of iron-phosphate complexes, which were formed by the reaction of the phosphate buffer in the eluent with the steel surface of the HPLC system. The effect of pH (3 to 10) on retention was examined, and phosphate was the only buffer found to be reasonably compatible with detection at 215 nm. Gradient elution was required to ensure removal of the iron phasphate which otherwise sorbed onto the column and ruined column performance. The presence of iron in the samples was a further complication due to precipitation of iron phosphate and the formation of iron complexes with the acids. Consequently this system was not studied further and, unless specified otherwise, the remainder of the discussions in this paper are limited to indirect detection systems. For indirect UV detection the absorbance of an eluent component (visualization reagent) is monitored, and the changes in ita concentration as a sample component is eluted can be related to the concentration of the sample component in the eluate (9-11). A particularly attractive feature of indirect detection for ion-exchange separations is that, in principle, it can be used for the detection of all anions without wide variations in sensitivity. The basic principles of the

817

detection mechanism (exchange of visualization-reagent anion for sorbed sample anion) are simple, but for successful application the interrelationships between eluent strength, eluent-anion charge, eluent concentration, background absorbance, and pH effects must be considered; with “permanently” sorbed exchange sites the effective ion-exchange capacity of the column can be varied to help optimize these interrelationships. In this study a number of visualization reagents were tested, and these are listed in Table I. All of these compounds can be used as eluents for indirect detection, but benzoic, benzenesulfinic, and salicylic were too weak (long retention times) and 1,2,4,5-benzenetetracarboxylic acid was too strong (short retention times) for separation of the multifunctional carboxylic acids studied here; benzoic acid was useful for separation of weakly retained carboxylic acids. The sensitivity of indirect detection depended on the charge on the eluent and sample anions (this determined the number of molecules of eluent anion exchanged for one sample anion), the molar absorptivity of the eluent anion, and the concentration of other anions in the eluent. The detection limits given in Table I1 (3 times peak-to-peak base line noise) illustrate the dependence of sensitivity on eluent charge and molar absorptivity (Table I). Anionic buffers (such as phosphates) or other anions in the eluent reduced sensitivity due to competition with the indirect-detection reagent for free ion-exchange sites when sample anions were eluted. Consequently a cationic buffer, THAM, was used, as it had no appreciable effect on Sensitivity. THAM was added to the acid form of the visualization reagent and thus was present as a salt of the visualization reagent. Column Systems. Both bonded-silica (C1J and styrenedivinylbenzene (PRP-1) reversed phases were used in these studies. The PRP-1 phase could be used at higher pH values, but the UV absorbance of OH- limited the usefulness of this feature. The C18 phases were more efficient, due to their smaller particle size, and the best performance was obtained with the 3 pm CISphase. Dynamic ion-exchange (ion pairing) separations were investigated on these columns with tetrabutylammonium salts, but this approach was unsuccessful due to the reasons discussed in “Detection”. Some of the eluent systems used for indirect detection with “permanently”coated columns are given in Table 111. Figure 1shows a separation that is typical of the different systems studied, and essentially the same system was used in field tests during reactor decontamination. Both PRP-1 and CIS“permanently” coated columns were used for periods of 3-5 months, and only small changes in column efficiency and retention times were noted. The “permanently” coated exchangers were superior to dynamic ion exchangers (ion-pair chromatography) for indirect detection because the presence of the dynamic modifier in the mobile phase can complicate detector response. The columns did not appear to be susceptible to poisoning, but if a problem was encountered the ion exchanger was stripped (S)and then recoated. The ability to vary the ion-exchange capacity of the columns was found to be important for adjustment of retention times as the background absorbance in indirect detection places limitations on the changes that can be made in eluent strength for any one eluent system. A number of the parameters affecting column selectivitywere examined, and these are discussed below. All of the eluents listed in Table I influenced column selectivity, due to ion exchange effects arising from changes in charge and concentration,and due to hydrophobic interactions on the reversed phase; the latter are discussed in more detail below. Table I11 lists capacity factors, k’, for some of the eluents used. For some of these eluents the effect of pH on selectivity was studied over the range of 2.7-9.3. All of the eluents listed in Table I11 became stronger eluents with in-

ANALYTICAL CHEMISTRY, VOL. 57, NO. 3, MARCH 1985

618

8

$

FeDCTA

2

RETENTION TIME lminl

Figure 2. Effect of acetonitrile on retention time. Experimental conditions similar to those given in Figure 1.

Table IV. Effect of Column Capacity on Retention Time” retention time, min

k’ values relative to oxalate

anion

20 mgb

80 mgb

20 mgb

80 mgb

oxalate FeDCTA EDTA citrate DCTA

1.9 3.3 5.3 6.0 9.8

2.6

1 7.1 12.3 14.2 24

1 0.4 4.1

2.0 8.9 10.9

15.0

5.0 7.0

“Eluent, 5 X mo1.L-l sulfoealicylate, 1.5 X mo1.L-’ THAM at pH 7.3; flow rate, 2.0 mL-min-’. bWeight of cetylpyridinium chloride sorbed onto PRP-1 column. creasing pH values, but the charge on some of the sample anions, such as EDTA and DCTA, increased with pH faster than that for the eluent anions. Consequently high pH values could not be used due to the long retention times for these acids. Some changes in selectivity with pH were observed, and this was a factor in the differences in retention time obtained between E and F in Table 111. Eluents with pH values below 5 were not useful as most of the sample acids were close to fully protonated at this pH; thus sorption was weak and similar for most of the acids, which made their separation difficult. For most of the compounds studied the optimum pH range was -7 to 8; pH 7.3 was chosen as a compromise between column stability and the buffering capacity of THAM. Hydrophobic interactions also influenced retention and the magnitude of this effect depended on pH, ion exchange capacity of the column, and solvent composition. Figure 2 shows that modification of the hydrophobic effects with acetonitrile could cause significant differences in relative retention. The retention of the more hydrophobic and lower-charged anions decreased with increased addition of acetonitrile to the eluent while the opposite effect was observed with the more polar anions. The hydrophobic effect is an attractive feature of these “permanently” coated reversed phases as it is another parameter that can be used to optimize separations; however, if more than -10% (v/v) acetonitrile is used the “permanently” sorbed ion exchanger may be slowly eluted. The ion-exchange capacity of the “permanently” coated columns can be changed quickly, and this can be also used to optimize selectivity. This is illustrated by the retention times given in Table IV for a PRP-1 column with two different exchange capacitites. The retention time for Fe”’DCTAdecreased unexpectedly with an increase in ion-exchange capacity. This decrease is likely related to the hydrophobic effects discussed above; the increased surface coverage of cetylpyridinium ions would reduce the hydrophobic surface area on PRP-1 and this would decrease retention for those ions where hydrophobic effects are important for sorption. Metal Ion Interferences. The presence of dissolved metal ions, particularly iron, is a common feature of chemical

cleaning procedures. For acids that function as multidentate chelating agents, well-defined metal-chelate peaks were often observed, and this was investigated as a basis for their determination. However, the addition of excess metal ion to the sample or to the eluent was required, and some of the excess metal ion usually sorbed onto the column (or precipitated onto the column) and caused badly skewed peaks for some of the acids which formed less stable metal chelates. This sorption of metal ions was particularly evident when salicylate eluents were used, and although it was much less important for phthalate eluents, the latter eluents also gave unsatisfactory results for the separation of a number of acids as metal chelates. Thus a different approach was examined, consisting of the addition of excess DCTA to the sample to complex all of the free metal ions, and subsequent analysis of the free acids. The effectiveness of this approach was first examined for iron, because it was the predominant metal ion in the sample, and was present in concentrations up to -250 ~ g . mL-’. Aside from forming complexes with the acids, dissolved iron introduced analytical errors due to redox, photochemical, and kinetic effects. Ferric complexes of multifunctional carboxylic acids can be photoreduced (1,12-16) to produce Fe(I1) and ligand oxidation products. This reaction caused low results when samples were not protected from light. FeII’EDTA- can also undergo a thermally activated reduction accompanied by oxidation of EDTA (17)but this should not occur with the experimental condition used for sample analysis (see Experimental Section). When dissolved oxygen was present, Fe(I1) was rapidly oxidized to Fe(III), and thus a cyclic oxidation-reduction occurred if light was present also. This cyclic reaction eventually destroyed the acids, and Fe(II1) precipitated due to a rise in pH. Consequently it is imperative that all samples be protected from light. Dissolved oxygen caused additional problems when it was present during the preparation of standards containing iron by the direct reaction between the acids and iron powder. Under these conditions 90% to 100% losses were observed for EDTA, and a peak for a decomposition product was observed. Good recoveries were obtained when standards were prepared under helium, but dissolution of the iron powder was much slower, which suggests that Fe(II1) or dissolved oxygen may affect the mechanism of interaction between these acids and iron surfaces. Standards prepared under helium were only faintly colored, and rapidly changed to green-yellow when exposed to air. The exchange of iron among the ligands was rapid only for Fe(I1). For Fe(III), however, especially for multidentate ligands such as EDTA, exchange is slow (half-life = 16 h at pH 2), and the half-life increases with pH (12). Consequently liberation of the free acids by the preferential complexation of Fe(II1) by DCTA was slow, and low recoveries were obtained. The effects of this slow exchange are illustrated by the data in Table V. Total elimination of oxygen was difficult, and even with a He purge low recoveries (70-90%) were obtained (sample 1). If the sample was allowed to stand prior to addition of DCTA (sample 2) larger losses were obtained due to oxidation of Fe(I1) from oxygen that diffused through the polyethylene container. Good recoveries were obtained if DCTA was added during dissolution of the standards (sample 3), but this cannot be used for the analysis of samples. However, reduction of the Fe(II1) by hydroxylamine hydrochloride (sample 4)promoted rapid exchange of the iron and good recoveries were obtained (sample 4). Investigation of this reduction procedure showed that less than 50% of the iron was complexed by DCTA after 1.5 h when reduction was done at ambient, but rapid (5-10 min) reduction and complexation by DCTA were achieved if the

ANALYTICAL CHEMISTRY, VOL. 57, NO. 3, MARCH 1985

619

Table V. Recovery of Acids in Presence of Dissolved Irona % sample recovery

anion

1 2 3 4 He purge He purge He purge NH,OH*HCl (new) (4-days old) (DCTA added) reduction

oxalate 106 (33)b 109 (33) 32 (44) EDTA 86 (44) 103 (33) citrate 110 (33) Feconcn (22) (22)

104 (60) 102 (80) 101 (60)

106 (100) 102 (116) 101 (67)

(30)

(33)

P

(Pp.

mL-') Experimental conditions: All samples were made 1000 pp.mL-' in DCTA prior to analysis; samples 1-3 were prepared by dissolution of Fe powder by the acids under He; DCTA was added to 3 prior to dissolution of Fe; the Fe in sample 4 was added as NaFe"' EDTA prior to reduction. All recoveries were calculated from response factors obained in the absence of Fe. bNumbersin parentheses are starting concentrations in Pg-mL-'.

I

I

I

I

I

2

4

6

8

TIME 400 r

(HOURS)

Flgure 4. Comparison of analytical results obtained during reactor decontamination: HPLC data indicated by 0, data from other techniques indicated by Experimentalconditions: HPLC results obtained with 10 pm C,* column under conditions similar to those for Figure 1; results for EDTA obtained by GC procedure ( I ) ; ion chromatographic separation for oxalate and citrate done by standard ion-exclusion procedure of Dionex Corp.

+.

300 4

B

;200 a " L 100

200

400

600

800

SAMPLE CONCENTRATION (pg.rnL-')

Flgure 3. Calibration curves for separation on a coated 3 pm column. Experimental conditions similar to those given In Figure 1.

samples were heated in a 100 "C water bath. The pH also affected this reaction and lower pH values (2 to 4) gave better recoveries. Small losses (5-7%) were sometimes observed if the samples were heated for 1 h or more, or when they were allowed to stand for 1day; for these samples small peaks for decomposition products were also observed. During these studies it was found that cap liners of the sample vials could cause significant loses of some acids when the samples were heated to 100 "C; this effect was eliminated by the use of Teflon-lined caps, and 100% recoveries were obtained for the acids tested (HEDTA, oxalate, EDTA, citrate, and DCTA). The effect of Ni(I1) and Cr(II1) was also studied briefly. Ni(I1) formed a stable complex with EDTA (this is probably true for the other multidentate ligands) but was quantitatively exchanged to DCTA after 10 min at 100 "C; recoveries for 350 pg-mL-l EDTA were 97% and 98% in the presence of 50 and 100 HgmL-' Ni(II), respectively. Cr(II1) slowly (few minutes) formed a nonlabile complex with EDTA and no exchange with DCTA was observed even after reaction at 100 "C for 30 min. As the Cr"'EDTA was eluted at the solvent front, EDTA bound by Cr(I1D could not be determined; this may be possible if weaker eluents are used. Analytical Performance. Figure 3 shows calibration curves obtained for separations on a 3 pm C18column. The curves were slightly nonlinear, and this was especially evident for EDTA at low concentrations, which was attributed to interactions of EDTA with metal surfaces or metal ion impurities. At the start of each day low results ( 10%) were observed for EDTA, and two to three injections were required to condition the system. Peak areas were used for quantitation, because indirect detection is an inherently nonlinear system, and peak heights, which are more sensitive to peak shape, usually gave calibration plots that had much smaller linear ranges. Reproducibilities (la) for the analysis procedure

-

were determined several times, and were always in the range of 3 to 5% for HEDTA, oxalate, citrate, EDTA, and DCTA in the concentration range of 75 to 350 pgmL-l; these data were obtained under conditions similar to those given for Figure 1. A complete separation (Figure 1)took -11 rnin (5-7 min at faster flow rates) and with a 5-10 min reaction with hydroxylamine and DCTA, total turn-around time was 15 to 21 min for any one sample. Any anion that is sorbed strongly onto anion exchangers is a potential interferent; of the common inorganic anions tested (F,Cl-, NO, NO2,Br-, and SO:-) only Sod2showed appreciable sorption and gave a positive interference for oxalate (10 pgmL-' of S042- gave a peak corresponding to -10 pgmL-' of oxalate). The pH of the injected sample affected peak shape for some of the acids and the size of the system peak; the magnitude of these effects depended on the eluent. Sample pH, which was studied over the pH range of 2.7-9.3, was unimportant for the sulfosalicylate eluent but was important for the phthalate eluent. For this latter eluent sample pH should be 7.3-8.0, especially for EDTA determinations; this pH adjustment is not difficult as the sample is buffered by THAM, and the same volume of base is usually sufficient for pH adjustment of all samples obtained during a decontamination. Figure 4 shows a summary of some of the results of obtained during tests of these techniques during reactor decontaminations; the data generated by the power-plant staff (Figure 4) were obtained by gas chromatography for EDTA (nitrogen selective detector) and by ion chromatography (Dionex) for citrate and oxalate. These results show that good agreement was obtained; the iron concentrations in these samples varied from 20 to 250 pgmL-l. This chromatographic system can also be used to determine metal ions simultaneously with the acids; such analyses are used to monitor corrosion. For iran, which will usually be the major metal ion present, this determination can be direct (measurement of the negative FeI'IDCTA- peak) or indirect (measurement of change in peak area for DCTA). For some of the chromatographic systems tested, inadequate resolution prevented direct measurement and the indirect measurement was used. Table VI shows a comparison of the indirect HPLC and atomic absorption (AA) results for a number of samples

620

ANALYTICAL CHEMISTRY, VOL. 57, NO. 3, MARCH 1985

Table VI. Comparison of Indirect HPLC and Atomic Absorption Results for Iron in Chemical Cleaning Solutions" found, ue.rnL-' HPLC AA F e concn

197 205 171 194 215 203 192 212 192 248 180 248 OAverage

178 197 185 190 191 202 180 196 180 220 170 230

ratio (HPLC/AA) 1.107 1.041 0.924 1.021 1.126 1.005 1.067 1.082 1.067 1.127 1.059 1.078

to adjust the relative retention time of this NiI1DCTA2chelate.

CONCLUSIONS Ion exchangers formed by the sorption of hydrophobic cations onto reversed phases can offer significant advantages over bonded phase exchangers for the separation and determination of multifunction carboxylic acids that are used for chemical cleaning of industrial components such as heattransport systems of nuclear-power reactors. The ability to optimize ion-exchange capacity and hydrophobic effects permits the development of separations that may be difficult on other types of ion exhcangers. This system has been proven to give reliable analysis under actual decontamination conditions and should be capable of replacing the two to three instruments used for such analyses in the past. Registry No. EDTA, 60-00-4; HEDTA, 150-39-0; citric acid, 77-92-9; oxalic acid, 144-62-7; iron, 7439-89-6.

of ratios = 1.059. Standard deviation = 0.057.

obtained during different decontamination tests. These data show that there was a reasonable agreement between the two techniques, but a "t test" showed that the average of the ratios was significantly different from 1.0 at the 99% confidence level. The presence of a significant positive bias for the HPLC results is supported by the fact that although HPLC calibration curves for iron were linear, they gave a positive intercept at the "Y" axis. A possible source of this positive bias may be the reaction of DCTA with metal surfaces or with other metal ion impurities in the HPLC system. Direct measurement of the FeII'DCTA- peak was preferable, especially for small iron concentrations, and linear calibration curves, which went through the origin, were obtained; detection limits were 0.4 pg.mL-'. Adjustment of the hydrophobic interactions via acetonitrile content of the mobile phase was useful for resolving FemDCTA- from the acids of interest. The Ni"DCTA2- chelate gave a well-defined peak (Figure l),and thus it should also be possible to determine the concentration of this metal ion; however, this was not investigated. It should also be possible to use the acetonitrile composition

LITERATURE CITED (1) Cassidy, R. M.; Harpur, R.; Elchuk, S. J. Chromatogr. 1980, 790, 188- 192. (2) Rudiing, L. Wafer Res. 1971, 5 , 831-837. (3) Chen, S. G.; Cheng, K. L.; Vogt, C. R. Mlcrochlm. Acta 1983, 473-48 1. (4) Perfetti, G. A.; Warner, C. R. J. Assoc. Anal. Chem. 1979, 62, 1092-1095. ( 5 ) Parkes, D.G.; Caruso, M. G.; Spradling, J. E., I11 Anal. Chem. 1981, 53,2154-2158. (8) Chinnick, C. C. T. Ana/yst(London) 1981, 106, 1203-1207. (7) Harmsen, J.; Van Den Toorn, A. J. Chromafogr. 1982, 249, 379-384. (8) Venezky, D. L.; Rudzinski, W. E. Anal. Chem. 1984, 56, 315-317. (9) Cassldy, R. M.; Elchuk, S. J. Chromafogr. Sci. 1983, 27, 454-459. (10) Small, H.; Miller, T. E., Jr. Anal. Chem. 1982, 5 4 , 462-469. (11) Hackzell, L.; Schlll, G. Chromafographla 1982, 75,437-444. (12) Jones, S. S.; Long, F. A. J. fhys. Chem. 1952, 56, 5-33. (13) Hill-Gottingham, D. G. Nature (London) 1955, 775 347-348. (14) Hall, L. H.; Lambert, J. L. J. Am. Chem. SOC. 1988, 9 0 , 2036-2039. (15) Trott, T.; Hemwood, R. W.; Lanford, C. H. Environ. Sci. Techno/. 1972, 6 , 367-388. (16) Natarajan, P.; Endlcott, J. F. J. fhys. Chem. 1973, 77, 2049-2054. (17) Motekaitis, R. J.; Martell, A. E.; Hayes, D. Can. J. Chem. 1980, 56, 1999-2005.

RECEIVED for review September 20,1984. Accepted November 27, 1984.