84
Anal. Chem. 1982, 54, 84-87
The di- and triphosphate adenosine nucleotides (ADP and ATP) had much longer retention times than the monophosphate nucleotides; ADP required 35-40 min while ATP was not removed from the column even after an hour. Guanosine monophosphate (GMP) was successfully analyzed in this system but could only be partially separated from CMP under the conditions employed. The detection limits and precision were somewhat degraded following separation on the columns. Although the sensitivity of the ICP system for nucleotides (based on phosphorus results) is less than that of the UV detector, it can provide valuable additional information as an LC detector. Applications involving compounds other than nucleotides may also benefit from the information provided by an ICP system.
LITERATURE CITED
Iyl; 1,;;
, rb
,
,
111 -
0 2 4
_I
v
~
\
10 1
14
Kratovich, M.; Roe, 8. A. J. Chromatogr. 1978, 755, 407. Brown, P. R. J. Chromatogr. 1970, 52, 257. Trultt, D.; Robinson, J. W. Anal. Chlm. Acta 1970, 49, 401. Windsor, D. L. Ph.D. Dissertation, Unlverslty of Arlzona, Tucson, AZ, 1977. (5) Windsor, D. L.; Denton, M. B. Appl. Spectrosc. 1978, 32, 366. (6) Boumans, P. W. J. M. ICP Inf. Newsl. 1979, 5 , 181. (7) Barnes, R. M. CRC Crit. Rev. Anal. Chem. 1978, 7 , 203. (8) Windsor, D. L.; Denton, M. B. J. Chromatogr. Sci. 1979, 17, 492. (9) Windsor, D. L.; Denton, M. B. Anal. Chem. 1979, 57, 1116. (10) Northway, S. J.; Fry, R. C. Appl. Spectrosc. 1980, 34, 332. (11) Northway, S. J.; Fry, R. C. Appl. Spectrosc. 1980, 34, 338. (12) Gast, C. H.; Kraak, J. C.; Poppe, H.; Maessen, F. J. M. J. J. Chromafogr. 1979, 185, 549. (13) Fraley, D. M.; Yates, D.; Manahan, S. E. Anal. Chem. 1979, 51, 2225. (14) Morita, M.; Uehiro, T.; Fuwa, K. Anal. Chem. 1980, 52, 349. (15) Windsor, D. L.; Heine, D. R.; Denton, M. 8. Appl. Spectrosc. 1979, 33, 56. (16) Heine, D. R.; Babis, J. S.; Denton, M. B. Appl. Spectrosc. 1980, 34, 595. (17) Fry, R. C.; Denton, M. 8. Anal. Chem. 1977, 49, 1413. (18) Fry, R. C.; Denton, M. 8. Appl. Spectrosc. 1979, 33, 393. (19) Garbarino, J. R.; Taylor, H. E. Appl. Spectrosc. 1980, 34, 584. (20) Gustavsson, A. ICP Inf. Newsl. 1979, 5 , 312. (21) Tliden, S. B.; Denton, M. B. J . Aufom. Chem. 1979, 3 , 129. (22) Skogerboe, R. K.; Grant, C. C. Spectrosc. Lett. 1970, 3 , 215. (1) (2) (3) (4)
I
TIME, mln
Flgure 5. Comparlson of a UV and an ICP detector for the separation of three nucleotMes using a 5-min linear gradient elution: solvent peak (a), nucleoside (b), AMP (c), CMP (d), UMP (e), and H,P04- (f).
nucleoside which does not contain a phosphate group. These two peaks are not present in the ICP recording (upper trace) because neither of these compounds contains phosphorus. Both detectors respond to the nucleotides AMP (adenosine monophosphate), CMP, and UMP, respectively, which correspond to peaks c, d, and e. The slight time lag between the UV and ICP recordings is due to the physical separation of the two detectors as previously described. Inorganic phosphate, presumably the result of partial hydrolysis of one or more nucleotides, is responsible for peak f in the ICP output demonstrating the advantage of two detectors.
RECEIVED for review January 6, 1981. Accepted September 25,1981. This work was partially supported by the Office of Naval Research and by an Alfred P. Sloan Foundation Fellowship awarded to M.B.D.
Solvent Extraction-Ion Chromatography for Determination of Chloride in Liquid Bromine P. F. Relgler," Norman J. Smith, and V. T. Turkelson Analytical Laboratories,
1602 Building,
The Do w Chemical Company, Midland, Michigan 48640
The productlon of bromlne by the "blow-out" process results In a relatively pure product (99.8%) whlch can still contain up to 2000 ppm chloride. A solvent extraction-Ion chromatography method has been developed for the determlnation of chlorlde in llquid bromlne. The sample is dlssolved in a potassium bromide solullon and the free bromine Is separated by extraction wlth carbon tetrachlorlde. The extracted aqueous solutlon is chromatographed on a column of low capaclty, pellicular anion exchange resin using conductivity detectlon to measure the chloride response. The method has a relative precislon of 3~2.3% at the 95% confldence limit.
In the "blow-out" process for the production of bromine,
the bromide in either acidic brines or seawater is oxidized to bromine by chlorine and the mixed halogens are blown out by either steam or air. The crude bromine is condensed to a liquid, passed through a separator, and finally purified in a stripper column producing a product of 99.8% purity (I). This relatively high purity product can still contain up to 2000 ppm residual chloride. Chloride and/or free chlorine is determined in bromine by a classical complexometric titration using mercuric nitrate as the titrant. A known volume or weight of liquid bromine is added to a potassium bromide solution which converts chlorine to chloride. The mixture is boiled to volatilize the free bromine. Subsequently, excess bromide is oxidized to bromine by the addition of a potassium iodate-nitric acid solution and the mixture is reboiled to remove bromine. The cooled so-
0003-2700/82/0354-0084$01.25/0 0 1981 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 54, NO. 1, JANUARY 1982
lution is titrated with. a standard mercuric nitrate soluti.on using sodium nitroprusside as a turbidimetric indicator (2, 3). This methodology h,as several drawbacks which make it less than desirable. Foremost among these is the need of a large volume capacity fume-scrubber hood capable of containing the evolution of 10-40 g of free bromine. Not only are such hood systems expensi.ve t o construct but they are costly t o maintain. Secondly, the turbidimetric end point is very difficult to perceive and duplicate. Finally, for accurate results, a titration correction vidue must be established by each analyst and must be subtracted from the volume of titrant used prior t o the subtraction of the blank. The total volume of solution and t h e concentration of the halide a.re the prime factors influencing this value; however, acidity, titration rate, and perception of the end .point can also contribute but t o a lesser extent. A new solvent extraction-ion chromatographic method :has been developed whidh does not require the use of a fumescrubber hood. In this procedure, bro:mine is solubilized in a potassium bromide solution and the free bromine is removed by extraction with carbon tetrachloride. The remaining aqueous solution is chromatographed on a column of low capacity, pellicular anion exchange resin ( 4 ) . Conductivity detection is used to monitor the eluted chloride. The method is applicable to the concentration range of 2-1000 ppm chloride and has a relative precision of 12.3% at the 95% confidence limit.
EXPERIMENTAL SECTION Reagents. Carbon tetrachloride (CC:I4)was used as the extraction solvent and wiw an analytical grade. Caution: carbon tetrachloride is a toxic c:hemical. Avoid inhalation of fumes and exposure of the eyes and skin to the liquid. Handle only in a well-ventilated hood. The potassium bromide solution was prepared by dissolving 0.200 g of reagent grad.€!potassium bromide (KBr) in deionized water in a 2-L vo1umet:riic flask. The flask: was diluted to volume with deionized water and mixed thoroughly. One milliliter of this solution contains 100 pg;of KBr. This solution was dispensed from a Brinkman, 0-25 mL Dispensette (Brinkman Instruments Corp.). A chloride standard solution (A) was prepared by dissolving 0.2478 g of reagent grmde sodium chloride (NaC1) in deionized water in a 500-mL vol.umetric flask. The flask was diluted to volume with deionized water and mixed thoroughly. One milliliter of this solution A contains 300 pg of chloride (Cl-). A working standard chloride solution (B), calculated to contain 2.25 pg of chloride (C1-) and 100 bg of KBr per millliliter, was prepared by adding 1.5 mL of standard chloride solution A to a 200-mL volumetric flask. The flask was diluted to volume with the potassium bromide soluiti.on listed above and mixed thoroughly. The eluent was an equimolar, 0.002 M sodium carbonate-sodium bicarbonate solution. This solution was prepared by dissolving 0.424 g of reagent grade sodium c;ubonate (Na2C03)and 0.336 g of reagent girade sodium bicarbonate (NaHC03) in deionized water in a 2-1, volumetric flask., The flask was diluted to volume with deionized water and mixed thoroughly. The 1.0 N sulfuric acid solution used to recondition the stripper column was purchased commercially. It can also be prepared by carefully and slowly adding 30 mL of concentrated sulfuric acid (H2S04),density 1.84 g/cm3, into 500 nnL of deionized water. Caution: corrosive chemical. Avoid exposure to the eyes and skin. The resulting solution is cooled to room temperature and transferred to a 1-L volumetric flask, dliluted t o volume with deionized water, and imixed thoroughly. Instrumentation. The ion chromatograph used was a precursor of an instrumeint, Model 10, available now from Dionex Corp., Sunnyvale, CA. The chromatographic conditions used are described below. The separation coliumn was a 450 mm X 3.0 mm i.d. glass column packed with ix low capacity anion exchange resin agglomerated on surface sulfonated styrene divinylbenzene available only from Dionex Corp. The stripper column or suppressor
85
E
A
C
Figure 1. Typical ion chromatogram; (A) sample injection point, (B) sample chloride response, (C) standard injection point, (D) standard chloride response, (E) and (F) bromide response of sample and standard injections, respectively. See the text of the Experimental
Section for chromatographic conditions. column was a 500 mm X 3.0 mm i.d. glass column packed with AG 5OWX16 cation exchange resin (Bio-Rad Laboratories, Richmond, CA.) in the hydrogen form (200-400 mesh). A 1 N sulfuric acid solution was used for column regeneration. The eluent was an equimolar, 0.002 M solution of sodium carbonate and sodium bicarbonate in deionized water at a flow rate of 1.5 mL/min. The injection volume was 100 pL. Detection was by electrical conductivity using a 2-pL flow cell. The recorder was operated a t 5-mV full scale and a chart speed of 30 cm/h. Cap liners made of Teflon and asbestos for the caps to the 1-oz Boston round glass bottles were obtained from Garlock Inc., Camden, NJ. Preparation of Sample. Twenty milliliters of the potassium bromide solution (100pg of KBr/mL) was transferred into a clean, dry 1-oz glass bottle equipped with a cap having a Teflon and asbestos liner. The bottle was capped and weighed to the nearest tenth of a milligram. With an Eppendorf, 200 pL, pipettor, 0.2 mL of liquid bromine was added to the contents of the bottle. Caution: liquid bromine and its vapors are toxic. Avoid exposure to the eyes and skin. Use only in a well-ventilated hood. The bottle was recapped immediately and again weighed to the nearest tenth of a milligram. The difference in weight between the second and the first weighing5 were the grams of sample taken. The bottle and contents were shaken vigorously by hand for 1.5 min. The solution was transferred to a 60-mL separatory funnel. No attempt was made to effect a complete transfer. The solution was then extracted as rapidly as possible with four 20-mL portions of carbon tetrachloride. The lower organic layer was discarded. In order to accomplish a good separation, a small amount of the aqueous phase was allowed to pass through the bore of the stopcock. The colorless,extracted, aqueous layer was transferred to a 4-dram vial equipped with a polyseal cap. The vial was shaken and allowed to stand to coalesce carbon tetrachloride droplets. Analysis Procedure. The 3-cm3 syringe was flushed twice with water and then the sample loop was flushed with water from the syringe. The syringe was then flushed with the sample solution and reloaded and the sample loop was charged with the sample. The point of injection was marked on the recorder chart paper. The chloride ion elution time varied from 5.9 to 7.1 min depending on the conditions of the particular column. Upon return to base line following the chloride response, the above procedure was repeated using the standard chloride solution B. The eluent was allowed to flow through the column for an additional 30-35 min to allow extraneous anions, namely, bromide, to elute. A typical chromatogram is shown in Figure 1. Quantitative results were obtained by the following calculation: ppm chloride (C1-) = A B V / C D where A = concentration of the standard chloride solution B
86
ANALYTICAL CHEMISTRY, VOL. 54, NO. 1, JANUARY 1982
Table I. Variations in Sample Weights When Pipetting 0.1 mL of Liquid Bromine (Density of Bromine at 25 "C, 3.11 g/cm3) mL of sample taken 0.1 0.1 0.1 0.1 theoretical sample wt 0.311 0.311 0.311 0.311 actual grams of sample 0.201 0.250 0.183 0.213 expressed in pg of Cl-/mL, B = peak height of the sample responses measured in millimeters, V = extraction volume, 20 mL, C = peak height of the standard response measured in millimeters, and D = grams of bromine taken.
RESULTS AND DISCUSSION The chemical reactivity and the very high vapor pressure of liquid bromine (400mm at 41 OC) presented an interesting problem in sampling and solvent extraction. Initially, sampling was done on a volume basis using a 100-pL Eppendorf pipettor. On the assumption that a consistent size sample was taken, the sample weight of bromine used was obtained by multiplying the volume times the density of bromine, 3.11 g/cm at 25 "C. Results obtained by this method of sampling showed poor reproducibility, -&20% relative (2a) at the 61 ppm chloride level. Because of this, several variations in sampling were tried and included the following: (a) volume sampling with an Eppendorf pipettor but inclusions of the tip into the extraction media; (b) direct syringing with a syringe equipped with a platinum needle; (c) indirect dispensing by weight using a dropping bottle; (d) sampling by weight with a gelatin capsule; (e) use of a glass ampule to sample by weight; (f) dispensing with the Eppendorf pipettor on a direct weight basis. Sampling variation (a) proved to be an improvement in technique. The results showed a relative reproducibility of f12.4% (2a) at the 61 ppm chloride level. Syringing (b) proved to be ineffective again due to the vapor pressure of bromine. Vapor pressure and density caused erratic results in the use of a dispensing bottle (c). The gelatin capsule (d) provided a good concept in containment but bromine was found to react with the gelatin. The glass ampule (e) corrected the reactivity problem associated with the gelatin capsule but was found to be very difficult to load and subsequently difficult to break under the partitioning solution. Dispensing on a direct weight basis (f) was found to be the most viable approach. Table I shows the variations in sample weight that can occur when sampling on a volume basis. The expected sample weight (volume X density) was 0.311 g instead of those listed. Sampling is now done by adding the bromine to a previously weighed 20.0-mL portion of a potassium bromide solution contained in a 1-oz bottle fitted with a cap containing a Teflon-asbestos liner. After addition, the bottle and contents are reweighed. The difference in weight is the grams of sample taken. The Teflon-asbestos liner was found necessary due to the reactivity of the bromine. The volume of potassium bromide solution was chosen such that up to 0.67 g of bromine would be soluble. The solubility of bromine at 25 OC is 3.35 g/100 g of water (2). The existence of free chlorine in liquid bromine has not been verified nor has another possibility, bromine chloride. Consequently the amount of potassium bromide necessary to convert chlorine to chloride is not known exactly. In the classical volumetric method, potassium bromide to chloride ratio of approximately 40 to 1 on a weight basis is maintained. Experimentation here has shown that consistent results can be obtained on a given sample with or without the addition of potassium bromide. Samples analyzed without the addition of potassium bromide show a bromide response equivalent to approximately 350 ppm KBr. The presence of bromide in the bromine leads one to believe that the only chlorine there is present as chloride (Cl-) and probably as hydrochloric acid. However, as a precautionary
Table 11. Stability of Solutions Extracted Four Times Followed by Separation of the CCl, Layer vol of I3r taken, mL date prepared and chromatographed chloride response, mm
0.05 0.10 0.15 7/18/80 7/18/80 1/18/80 14
date reanalyzed chloride response, mm
7/21/80 7/21/80 7/21/80 16 21 26
19
24
measure the volume (20 mL) of aqueous solution chosen for partitioning is made up to contain 2000 pg of KBr total. Separation of the free bromine dissolved in 20.0 mL of 100 pg of KBr/mL solution was done with carbon tetrachloride (CC14). Several other solvents were also evaluated for extraction efficiency. They include toluene, o-dichlorobenzene, and chloroform. Toluene extraction capabilities were good; however, an emulsifying effect made interface separation difficult. In addition, toluene being less dense than water becomes the surface phase making repeated extractions cumbersome. o-Dichlorobenzene extracted bromine well but left the aqueous layer with a turbidity of unknown origin and composition. The extraction efficiency of carbon tetrachloride and chloroform was similar. Carbon tetrachloride was chosen because of the density of the solvent. Initially, only two CC14 extractions were used to separate the bromine. A 5-mL aliquot of CC14was used to extract the bulk of the bromine from the KBr solution. In the second extraction, 10-mL portions of the aqueous phase were extracted with 20 mL of CClk The lower CC14phase was pale yellow. The extracted aqueous phase was water white and used for ion chromatography. No effort was made to separate these phases. Results indicated that such a mixture was unstable. Upon being allowed to stand for periods in exceas of 2 h, the bromine remaining in the CCll appeared to substitute for the chloride resulting in abnormal chloride responses in the aqueous layer. T o correct this situation, we extracted the bromine containing KBr solutions four times with 20-mL aliquots of C C 4 , discarding the CC14layer after each extraction. See Table 11. The final extraction produced water white clarity in both phases prior to separation. Results on these solutions showed excellent stability and reproducibility. A potassium bromide solution containing no bromine was extracted as above. There was no blank contribution from this solution during chromatography. Similarly, the responses of an extracted and an unextracted chloride standard solution were identical. Therefore preparation of a reagent blank and extraction of the chloride standard solution were unnecessary. Column technology, choice of eluent, and detector were all studied earlier by Turkelson and were well-defined (5). There existed only the need to check out possible interferences due to sampling and extraction. Fortunately, the chromatogram proved to be free of interferences. A time span exists between the measurement of the chloride response and the elution of the bromide peak such that the calibration standard can be injected immediately following the chloride peak; see Figure 1. Research has shown that for the best possible results, chloride response (measured by peak height) should be at a maximum for the range of the chart paper and the standard and sample should be adjusted to elicit the same magnitude of response. Recovery. The recovery was measured by using the spiking technique. Two individual bromine samples, 13 ppm C1- and 180 ppm C1-, were spiked with varying amounts of ,chloride standard, extracted, and chromatographed. Results in Table I11 show that an average recovery of 100.8% was determined on the 13 ppm sample and an average of 98.5% on the 180 ppm sample.
Anal. Chem. 1082, 54, 87-91
Table 111. Recovery Study Using the Standard Addition Technique sample chloride, ppm chloride added, f i g chloride found, p g chloride in sample
1 13 15.1 23.0 7.8
1 13 30.2 38.2 7.7
2 1880 90.6 1846.6 102.6
2 2 180 180 151.0 181.2 253.7 278.2 100.9 103.1
taken, p g chloride recovery, p g 15.2 30.5 84:.0 152.8 184.1 % recovery 100.7 101.0 92.7 101.2 101.6 average 100.8 98.5
Precision. Nine independent samplings of bromine were analyzed, five on one dily and four on the next. By use of the values 179, 183, 176, 1.81, 182, 180, 180, 180, and 178, a standard deviation of !2.1 ppm a t the 180 ppm chloride con-
87
centration level was calculated. The relative precision at the 95% confidence level is *2.3%. LITERATURE CITED (1) Hampel, C. A. "The Encyclopedia of The Chemical Elements"; Reinhold: New York, 1968; pp 81-89. (2) Kolthoff, I. M.; Elving, P. J. "Treatlse on Analytical Chemistry"; Interscience: New York, 1961; Part 11, Voi. 7, Section A. (3) Remy, H. "Treatise on Inorganic Chemistry"; Elsevler: New York, 1956; Vol. I,Chapter 17. (4) Small, H.; Stevens, T. S.; Bauman, W. C. Anal. Chem. 1975, 47, 1801. (5) Turkelson, V. T. The Dow Chemical Co., Midland, MI, private communlcation, 1980.
RECEIVED for review July 17,1981. Accepted October 9,1981.
Multiple Solvent Extraction System with Flow Injection Technology Dennis C. Shelly, Thomas M. Rossl, and Islah M. Warner" Department of Chemistry, Texas A& M University, College Station, Texas 77843
A three-stage extractlon procedure for Ithe isolation of polycyclic aromatic compounds from compllcated sample matrices has been automiated by use of flow injection technology. Three slngle-step liquid-liquid extractlons are llnked together by multichannel1 pumping and resampling. In addltlon to the multiple extraction capability, the system demonstrates two other novel features. First, both Teflon and glass extraction coils are useld to mlnimlze sample carryover and memory effects. Second, microprocessor-controlled pneumatically actuated valvlss control sample lnjectlon and effluent concentration. The performance of the system is evaluated by high-performance Iilquid chromatography and video fluorometrlc analyses of both automated and manually performed extractlons of a crudel oil-ash resldue sample. The threeextraction system is rapid, reproduclble, and quantitative as compared to an identlcal manual proceldure.
There is an increasing need for specialized instrumentation for automated sample preparation. Three of the more important reasons for this realization are the inherent speed, precision, and interfacing capabilities of automated procedures. For routine analyses, one frequently encounters (a) large sample loads, (b) introduction of bi,w or error associated with performing a given technique, and (c) acquisition o i a miriad of miscellaneous laboratory equipment. The virtues of automated sample preparation and automated analyses, in general, are discussed in a monograph by Foreman and Stockwell (1). To date, the emphasis on laboratory automation has focused on the development oif more efficient analysis systems. An example is the video fluorometer, which greatly speeds analpis time by implementation of a novel optical system, multichannel detection deviice, and computerized data treatment (2,3).A result of these innovations is that sample preparation, particularly for a very complex matrix, is often the time-lim-
iting step for the determination of one or more fluorophores. A very effective use of automated sample preparation would be either direct or indirect coupling to a sophisticated instrument such as the video fluorometer. In this way both routine and research oriented applications of multicomponent fluorescence determinations would be greatly expanded. Both segmented flow analysis (SFA), also called continuous flow analysis (CFA), and flow injection analysis (FIA) have been employed for the automation of many types of chemistries. Several recent reviews enumerate the many applications for which SFA (4-6) and FIA (7-9) methodologies were utilized. Both techniques, SFA and FIA, have been adopted in the automation of simple liquid-liquid extractions. Since the initial work of Karlberg ( I O ) , a great many applications of FIA automated solvent extraction have appeared in the literature (11-16). The rapid development of automated solvent extraction is perhaps due to the frequency and importance of liquid-liquid extraction as a sample preparation technique. Several extraction procedures have been proposed for environmental analyses. Two schemes can be cited for the isolation of polynuclear aromatic compounds (PNAs) in fly ash extracts (17, 18). Additionally, a method was reported for the extraction and determination of individual organic compounds, including polynuclear aromatics (PNAs), from shale oil (19). These procedures are characterized by incorporation of multiple extractions which, until now, have not been successfully automated by FIA methods. When utilizing such rapid techniques as video fluorometry (VF), high-performance liquid chromatography (HPLC), and a combination of the two (HPLC-VF) for the determination of carcinogenic species in shale oil, it became evident to us that automated sample preparation would be beneficial for routine and research oriented investigations. This need in addition to our previous experience with the dimethyl sulfoxide (Me2SO)/pentaneextraction of Natusch and Tomkins (18) led us to automate this relatively useful multiple extraction procedure.
0003-2700/82/0354-0087$01.25/00 1981 American Chemlcal Society