1 and 2 or calculation of their Y intercepts shows a possikile maximum sensitivity of 4.6 X lo--% gram of potassium cyanide and 2.5 x 10-9 gram of sodium sulfide. The direct responscb of the gold anode to cyanide is indicative of the formation of a gold-cyanide complex, giving an increase in current* This current response slowly decays over several
minutes, and this decay may be attributed to the cathodic reduction of oxygen a t the platinum electrode. LITERATURE CITED
(1) Baker, B. B., Morrison, J. D., ANAL.
CHEM.27, 1306 (1955). (2) Gawron, O., Fernando, J., Chem. SOC.83, 2906 (1961).
J. Am.
(3) Gray, A. G., “hfodern ElectroplatP*252, Wiley, 1953(4)ing~’, McCloskey, J. A.,New ANAL. CHEM. 33, 1842 (1961). (5) McPhee, J., Cecil, R., “Advances in Protein Chemistry,” XIV, pp. 255-389, Academic Press, Sew York, 1959. (6) Miller, G. W., Long, L. E., George, G. M., ANAL.CHEM.36, 1143 (1964). RECEIVEDfor review May 13, 1963. Resubmitted January 24, 1964. Accepted February 7, 1964.
Amperorrietric Titration of Chromium in Mixtures of FIuoride Salts R. F. APPLE and H. E. ZITTEL Analytical Chemistry Division, Oak Ridge Nafional laboratory, Oak Ridge, lenn.
b A rapid and precise method for the amperometric titration of chromium in mixtures of fluoride ! ; a h has been developed. The salt is dissolved in a mineral acid mixture, after which the chromium is oxidized to Cr(VI) with argentic oxide. The Cr(V1) is then amperometrically titrated with ferrous sulfate. This reduction causes no change in cell current until the end point is reached where the current decreases. The titration is followed b y means of a polarograph equipped with a titration cell which consists of a pyrolytic graphite electrode vs. the S.C.E. The optimum potential for the titration is +l.O volt vs. the S.C.E. In the range of 1 tcl 50 pg. of Cr(VI) titrated, the relative standard deviation is less than 276. Of the interferences studied, cmly Mn(VII) and Ce(lV) interfere with the method a t the level of concentrcitions studied.
1”
THE operation o f the molten salt reactor, it is anticipated bhat chromium will be used as a monitor of corrosion in the nuclear reactor fuel (IO). I n past work with nonirradiated fuels, chromium has been the most abundant corrosion product and the most readily corrodec! of the elements in Inor-5 containers. I n response to oxidants or reductants which may be added to inhibit corrosion, chromium responds more rapidly and over a wider concentration range than iron or nickel. The most widely used colorimetric method for determinag low concentrations of chromium, based on the reaction of chromium(V1) with diphenylcarbazide, is sensitive; however, approximately 30 minutes are required for the analysis. Zit1,el ( I S ) and Lefort (7) reported that when aqueous solutions of chromium arc subjected to high levels of radiation, a parbial reduction
of chromium(V1) to chromium(II1) occurs. This radiation effect, therefore, necessitated the development of a method, more rapid than conventional procedures, which could be applicable to the determination of microgram quantities of chromium in solutions of irradiated nuclear reactor fuels. The first approach was to investigate the determination of chromium as Cr(II1) by spectrophotometric measurements. Several chromogenic reagents have been reported (1, d , I d ) , but the data given in the literature for these methods indicated-that sensitivity in the microgram range of concentration was lacking. The reaction of Cr(V1) with ferrous ion has been widely studied (4, 6). Visual titration methods, however, required more time than was desired and high precision was difficult to achieve in lower concentrations. The theoretical requirements for an amperometric titration of Fe(I1) with dichromate have been thoroughly covered by Lannoye (6) and others. Many investigators have shown that a very high sensitivity is possible with a n amperometric titration. Keily, Eldridge, and Hibbits (3) used amperometry to obtain a high degree of precision for the determination of the These purity of ferrous sulfate. methods were not entirely applicable because of interferences which are contained in a mixture of fused fluoride salts (LiF-BeFrZrF4-UF4, 65-29-5-1 mole yo). Miller (9) recently described advantages of the pyrolytic graphite electrode in potentiometric titrations. By substituting the graphite electrode in the titration assembly, we have investigated the amperometric end point for the titration of Cr(V1) with very dilute solutions of ferrous ion. These studies have shown that well defined and reproducible end points are
obtained with reagent concentrations appreciably lower than those required to give a satisfactory end point by other means. Concentrations on the order of 0.1 pg. per ml. of Cr(V1) can be titrated with a relative standard deviation of less than 2%. EXPERIMENTAL
Apparatus. A buret assembly which consists of a Harvard infusion pump (Catalog S o . 1100), 2 r.p.m., equipped with a 1-ml. syringe with a long offset tip calibrated to deliver 0.036 ml. of titrant per minute. Polarograph, O R S L Model Q-1160, chart speed, 10 divisions per minute. Conventional titration cell with a pyrolytic graphite electrode and a reference S.C.E. salt bridge. Sample solutions were stirred magnetically with Teflon-coated stirring bars. Reagents. Primary standard dichromate solution (0.1000M). Transfer exactly 2.942 grams of KBS No. 136 or equivalent primary standard K2Cr20, to a 100-ml. volumetric flask. Dissolve in 0.5V &Sod, dilute to the mark, and mix. Dilutions of this stock solution were used throughout this work. Ferrous ammonium sulfate hexahydrate. Transfer 0.7 gram of ferrous ammonium sulfate to a 100-ml. volumetric flask. Dissolve in 0.5X H2S04, dilute to the mark, and mix. Standardize the solution according to the recommended procedure. Argentic(I1) oxide, Divasil (obtained from Merck and Co.). Mineral acid mixture, 15 ml. of concentrated H2S04,15 ml. of concentrated “03, 5 ml. of concentrated HC1, and 1 gram of H3B03. Procedure for Dissolution of Sample. Transfer a portion of the salt, usually 2 grams weighed to the nearest 0.5 mg., to a 250-ml. beaker. Add 30 ml. of the mineral acid mixture to dissolve the salt. Heat to strong fumes of SOI. Cool, rinse the sides of the beaker with water, and VOL. 36, NO. 6, M A Y 1964
983
reheat to fumes. Cool, transfer t o a 200-ml. volumetric flask, dilute t o the mark with water, and mix. Procedure for Titration of Sample. From the sample, pipet into the titration cell a test portion t h a t contains 5 t o 50 fig. of Cr. Dilute to about 30 ml. with 0.531 HzS04. Add about 20 mg. of Ago t o oxidize the Cr to Cr(VI), then heat for 5 minutes t o reduce the excess oxidant (8). Cool, then fix the cell to the titration assembly. Deaerate 10 minutes with argon. Add titrant, 0.001V Fe(II), by turning the infusion pump switch to the drive position. Follow the progress of the titration by means of a polarograph, ORNL Model Q-1160 or equivalent, set a t 1.0 volt respective to the S.C.E. and appropriate current range. Continue adding titrant until a sharp deflection in the titration curve occurs. Determine the end point of the titration by extrapolating the two segments of the titration graph to their point of intersection. Establish the inches of chart travel equivalent t o 1 fig. of chromium by titrating a minimum of five aliquots from a standard solution of chromium. DISCUSSION
An amperometric titration with Fe(I1) was carried out at an indicator electrode voltage of -0.1 us. S.C.E., the potential being on the plateau of the polarogram of Cr(V1). Precise and reproducible results were obtained in titrating dilutions of a standard solution of Cr(V1). However, when Cr(II1) was oxidized with argentic oxide (S), no titration curve could be obtained because Ag(1) is also active at this potential. Other oxidants, such as peroxysulfuric acid with CoSO4/NiSO4 (3, 11) were investigated, but were unsatisfactory because of low results or interference a t this potential. Since the reduction of Cr(V1) with Fe(I1) at an
Table I.
Precision of Replicate Titrations
(Five replicates) Chromium, pg. Rel. std. Taken Found dev., % 10.0 20.0 30.0 40.0
10.1 19.8 30.4 39.7
1.0 0.7 0.9 0.8
Table II. Determination of Chromium in LiF-BeF2-UF4-ZrF4 Fused Salt
Chromium taken," pg.*
Rel. std. dev., 70
1.03 1.3 5.15 1.1 20.0 0.7 41.2 0.8 a Average of four replicates. * Determined from average of 5 aliquots analyzed by diphenylcarbazide method.
984
ANALYTICAL CHEMISTRY
z I-
Lz 3 Lz
\
Potential of pyrolytic graphite electrode, + 4.0v. VS. S.C.E.
Sensitivity 0.5 po.fulI scale [ F&I 5 x M 0'6 18.4 7
Ib-.
1
1
0.34
0.68
I
I
I
I
1.02
1.36
1.70
2.04
T I T R A N T , mi.
Figure 1. Typical curve for amperometric titration of Cr+6 with Fe+2
indicator electrode voltage of -0.1 us. the S.C.E. was impractical because of interference from the oxidant, the titration of Cr(T'1) a t a potential significantly more positive was investigated. Several titrations were made, to determine the most suitable voltage for the amperometric determination of Cr(V1) with Fe(I1) in 0.5M HZSO4. A positive indicator electrode voltage of 1.0 us. the S.C.E. was chosen for the titration. At this voltage the titration is as precise as the more negative voltages and is free from interference from foreign ions (Figure 1). The current decreases at the end point rather than increasing, as is normally the case. In this titration the decrease in current is due to the fact that the end point is reached when sufficient excess of Fe(I1) is present to show an anodic reaction. The maximum precision of the method was found when at least 30 chart divisions were covered by the titration. This range can be reached either by adjusting the aliquot size or by changing the strength of the titrant. Equal precision using either 0.01 or 0.001M Fe was attained. Table I shows the precision which may be expected from the method. The dilutions of the standard solution of chromium were carried through the argentic oxide oxidation and 0.00151 Fe was used as titrant. The relative deviation at all limits of Cr(V1) studied was 1% or less. Table I1 shows the accuracy which may be obtained. A 1-gram portion of the salt was dissolved and diluted to 100 ml. with 0.5-If HzS04. The chromium content of this salt was determined by averaging the results obtained from samples analyzed by the diphenyl carbazide method. The aliquot size was varied to determine the accuracy of the method over a wide range of chromium concentrations. K h e n the aliquot contained 1 pg. of Cr, O.OOOld1 Fe(I1) was used; 0.001X Fe(I1) was used for the other titrations. The data show that an accuracy of about 1% may be obtained for chromium concentrations as low as 1
p.p.m. Extreme care should be taken to obtain oxygen-free solutions for the lower concentrations. It will be noted that 0.5M H2S04 media were used in all cases. Since the procedure was developed for use in the hot cell facility, a precision and accuracy study determining chromium in nonirradiated fuel was made by remote operation. The analytical chemistry mockup cell designed to simulate hot cell conditions was used for each series of determinations. Aliquots were transferred by means of a remotely operated pipet. -411operations necessary to complete the titration were performed remotely. A series of five replicates was titrated daily until 20 determinations had been made. When the entire procedure was carried out in the simulated hot cell facility, a relative standard deviation of 1.5% was obtained for the determination of chromium in sulfuric acid solutions of fused fluoride salts (Table 111). The various possible interfering ions which could be considered unreactive with Cr(V1) were studied (Table IV). Of the cations studied, only Mn(VI1) interfered seriously with the titration, while Ce(1V) interfered to a lesser extent. Fortunately, manganese seldom occurs in the fused fluoride eutectic and when present rarely exceeds 5 p.1i.m. Therefore, it does not present a problem in this particular application. The reaction of Ce(1V) with Fe(I1) was investigated. The reduction of Ce(1V)
Table 111. Determination of Chromium in LiF-BeFz-UF4-ZrFd Fused Salt under Simulated Hot Cell Facility
Chromium taken," Day 1 2 3
4
Rel. std.
pg.
dev., yi
18.4 18.4 18.4 18.4
1.5 1.6 1.5
1.5
Average of 5 replicates. Table IV. Interference Studies on Amperometric Titration of 40 pg. of Cr(VI) with Fe(ll) Chromium present, 40 pg.
Cr Quantity, found, Cation Li +
Be +* Zr +4
Th+4 L- +e NI +z Fe + 3
Ce +? Mn+7 Rlo+6
mmoles 4 4
4 4 0 0 0 0 0 0
1 1 5 0001 001 01
pg.
39 40 39 40 40 40 40 43 82 40
6 0 7 4 6 0 2 0 0 4
Recovery,
7c
99 100 99 101 101 100 100 107 205 101
0 0 3 0 5 0 j
5
0 0
under the conditions g;iven for the titration mas found to proceed a t a much slower rate than the reduction of Cr(V1). While some interferenx occurs, it is not too serious over the range of concentration tested. During .:he normal oxidation of Cr(II1) to Cr(V1) with Ago, the U(IV) is also oxidized to U(VI), a t which oxidation state the uranium is not titrated by the Fe(I1). A pyrolytic graphite electrode was used throughout this ivork. KO poisoning of the electrode cccurred over two months’ operation. Since this method is designed to be operated remotely, the graphite has an advantage over the platinum electrode, because it is more inert to oxidizing agents. The titrations are performed a t a potential much more positive than the point where the
dropping mercury electrode begins to oxidize. I n general, the method is markedly free from interference under the conditions studied. No significant bias is present and the method is simple and straightforward.
(6) Lannoye, R. A., ANAL. CHEM. 35 , 558 (1963). (7) . , Lefort. M. J.. J . Chim. Phus. 54, 782 (1957). (8) Lingane, J. J., Davis, D. G., A nal. Chzm. Acta 15, 201 (1956). (9) . . Miller, F. J., ANAL. CHEM.35, 929 (1963). (10) Oak Ridge National LabEratory, U. S. At. Enersv Comm.. MSRP
S e m i k u a l Prog.xep.,” USAEC Rept.
LITERATURE CITED
( 1 ) Boef, G. D., DeJong, W. J., Krijn, G. C., Poppe, H., Anal. Chim. Acta 23, 557 (1960). (2) Karnaev, N. A., Levin, A. I., Katovskaya, N. L., Ind. Lab. 28,573 (1962). (3) Keily, J., Eldridge, A., Hibbits, J. O., Anal. Chim. Acta 21, 135 (1959). ~I
(4) Kolthoff, I. M., Nightengale, E. R., Jr., Ibzd., 17, 329 (1957). (5) Laitinen, H. A,, “Chemical Analysis,” ChaD. 17. McGraw-Hill. New York. 1966.
ORNL-3419, 119-20 (1963). (11) Savichev, E. I., Iskihakova, E. I., Flvazhnikova, L. F., Ind. Lab. 28, 433 (1962). (12) Sherif, F. G., Oraby, W. M., Sadek, H., J . Inorg. N z ~ l Chem. 24, 1373 (1962). (13) Zittel, H. E., ANAL. CHEM. 35, 329 (1963).
RECEIVED for review December 9, 1963. Accepted February 7, 1964. Work done a t Oak Ridge National Laboratory, operated by Union Carbide Corp., for the U. S. Atomic Energy Commission.
Effect of Column Length on Resolution under Norm a Iiz e d T im e Co nd iti ons BARRY L. KARGER’ and W.
D. COOKE
Baker Laboratory, Cornell University, Ithaca,
b Time normalization in gas chromatography is a technique in which the elution time is fi)ied b y adjusting the operating temperature of the column. The effect of different operating parameters on resolution has been studied under these conditions and contrasted to isothermal operation, in which time Is ignored. This study focuses particular attention on the change in resolution as column length i s varied and shows that an optimum length exists for maximum resolution. Short columns can give better resolution in shorter time and a t lower temperature than very long columns.
M
studies havc: been made of the effect of column parameters on the number of theoretical plates and the rrsulting change in resolution (3, 11, 13). These studies have usually been run isothermally and no control is exercised over retention time as conditions are changed. Time, however. can be important in all types of gas chromatography separations. Given an infinite amount of time, thporetically any separation can be made on any column and under any given set of conditions. One column ANY
Present address, Department of Chemistry, Northeastern University, Boston, illass.
N. Y.
may separate two solutes in 1 hour; another, in 3 hours. The column separating the solutes in 1 hour is better, even though the resolution is the same. Thus, evaluation of columns or column parameters depends on retention time, and confusing and conflicting conclusions appearing in the literature about the merits of columns may be due in part to a disregard of this parameter. I n addition, retention time may be important when a separation has to be performed with a specified time interval-for example, during on-stream analysis or because of rates of decomposition and isomerization. Recently, attention has been focused on retention time in studies of minimum time analysis ( I , 6, IO, 11, IS). Here a resolution is specified, and parameters are examined for the optimum conditions-i.e., minimum time for the specified resolution. Although minimum time analysis is valuable, two problems may arise: Sometimes the minimum time is short (11), perhaps a matter of seconds, and operational difficulties may consequently occur (1). The effect of a parameter on resolution is difficult to ascertain. Operation a t H,;, and examination of the effect of various parameters on the analysis time have been suggested (1). This procedure, however, while removing the first problem, still leaves the second. Retention time has thus been considered from two extreme vien-s-one
giving no concern to time, the other treating time as the dependent variable for quickest separation. A useful way of incorporating the time element, recently suggested ( 5 ) , is time normalization, in which the temperature is adjusted to make the last component of a mixture elute a t the same time with each column or column parameter change; thus retention time rather than resolution is normalized. as in minimum time analysis. The normalized times may be selected to avoid both very short time and long time operational difficulties. &o changes in resolution with changes in the parameter under study may be seen directly. Time normalization can also be used to find ways to increase resolution without an increase in time. When a greater resolution is necessary for the analysis, various parameter changes can be made: the particle size of the solid support or column temperature can be decreased, the column length can be increased, etc. However, these changes cannot be made without increasing the retention time. On the other hand a lower retention time, produced by use of a shorter column length, etc., will decrease resolution in isothermal operation. If, however, two parameters are simultaneously changedfor example, a shorter column is used and the temperature is then lowered to normalize the time-a gain in resolution VOL 36, NO. 6, MAY 1964
985
