The water content is given by: [(G G o b - d,/da. Gan'lT % Hz0 = 10Vd.G
+
50
+
LO.
20
1,
101
0 7.15
1 220
' 225
230
735
740
n (cc.) Figure 1 . Transition time, 7,as a function of volume of added KF solution
because the bigger part of the reactant is added a t the very beginning of the titration. I n other words, in the titration of two different water-containing samples, the time for reaching a point which differs from the inflection point by a definite small amount is practically the same. Let n and n' be the titrating volumes a t the inflection points corresponding to the titration of the sample and of the blank, respectively; G and G, the weight, in grams, of the salt and of the acetic acid in which it is dissolved; T,the titer of the Karl Fischer solution (expressed in mg. of HzO per cc.) ; V the volume of acetic acid and of the acetic acid solution introduced in the cell; and d, and d, the respective densities.
since d, = d, 0.0074b, where b is the percentage of salt in the acetic acid solution and d, = 1.051 a t 20" C. Some of the results obtained are reported in Table I. The results are independent of salt concentration, which means that the extraction of the water contained in the sample is complete. The volumes of acetic acid solution and of the blank were of the order of magnitude of 5 to 10 cc.; T was 1.5 to 2 mg. of HzO per cc. The same KHF2 samples mere analyzed by electrolytically generated iodine after dissolving in a salicylic acid solution in the Karl Fischer reagent. I n this case, account was taken of the rate of esterification by measuring the current necessary to maintain a constant voltage across the platinum voltammetric-control microelectrodes. The results are in agreement with those obtained with the acetic acid method, although the coulometric method does not permit a routine analysis, since the time required for dissolution is too long (1). The separation between the anodic and cathodic compartments of the cell is accomplished with an anionic exchange membrane (BDH-Permaplex A-20). Finally, we measured the limiting current relative to the evolution of hydrogen on a bright platinum cathode
Table 1.
Analysis of Four KHF Samples
Water found, Sample 1
2
3 4
Run 1 2 3 4 5 6 7 8 9 10 11 12 13 14
b 2.1 2.4 2.8 1.8 1 .6 ~. ~
1.3 1.3 1 .o 1.2
%
0.657 0.645 0.648 0.649 0.788
0.793 0.776
1.6 0.9
0.310 0.330 0.310 0.309
1.2 2.8
0.040 0.036
1.6
0.030
immersed in a melted bath of KHF2 a t 250" C. The values obtained from galvanostatic measurements are approximately proportional to the water content as determined by one of the two abovedescribed methods (2). LITERATURE CITED
(1) Barbi, G., Pizzini, S., High Tempera-
ture Chemistry Section Laboratories, CCR Euratom, Ispra, Italy, unublished data, 1961. (2P Pizzini, S., Barbi, G., Sternheim, G., High Temperature Chemistry Section Laboratories, CCR Euratom, Ispra, Italy, unpublished data, 1961. GIOVANNI B. BARBI SERGIOPIZZINI Chemistry Department Euratom C.C.R. Ispra, Italy
Vacuum Output Gas Chromatography SIR: The operation of the output of a capillary chromatographic column in vacuum has not yet been evaluated from the viewpoint of its analytical value. Such technique was not considered advantageous in either its experimental or theoretical aspects ( 3 ) . For example, the linear gas velocity in a column operating a t 40-p.s.i.g. input and 1p.s.i.g. output pressures increases only in a ratio of 1 to 3; while in the case of a column having the same input but vacuum (IO+ torr) output pressures, it will increase in a ratio of 1 to 10'. Recently Giddings (1) showed in his theoretical paper that in case of capillary columns using vacuum output and a carefully selected input pressure, a high speed separating system could be obtained. While-operating the output of a standard Golay column in vacuum (10-3 torr) the input of the column can be operated a t or above atmospheric 410
ANALYTICAL CHEMISTRY
pressures. Such pressure differences can be held off by the capillary column itself without using additional restrictions and valves. I n a previous paper (3) describing a vacuum type quantitative and qualitative (QQ) detector, we suggested the possibility of coupling the outlet of a capillary column directly into the input of the detector tube and operating the output of the capillary, accordingly, a t low pressures. The feasibility of such a system has now been investigated, and experiments have been made to evaluate the performance of such an arrangement. The experimental system is shown schematically in Figure 1. The carrier gas was helium. The sampling system consisted of a standard Perkin-Elmer type gas sampling valve ( S V ) ,a rubber septum injector port (SZ), and a split (8).I n some of the experiments the sampling system was connected directly to the input of the column, not utilizing the split. The capillary column (C) was a 150-foot, 0.010-inch i.d. stainless
steel tube coated on the inside wall with Dow Corning silicone high vacuum grease. A conventional high vacuum system pumped the QQ detector tube, and through this the column, continuously. Both the QQ detector tube and the vacuum system were described in a previous paper ( 3 ) . The column output pressure was monitored by the QQ detector tube by switching the ionization potential from 22 volts to 150 volts, ionizing the helium gas and measuring its pressure. The input pressure was measured by a manometer (MI. The performance of this vacuum output chromatograph system was compared to a conventional system operating a t atmospheric output pressures and using a flame ionization detector. The measurements were carried out using identical columns, identical input pressures, sample amounts, and operating temperatures. The gas mixture to be tested contained CH4, C2H6, and
CAPILLARY COLUMN
00 OETECTOR TUBE
li
I!
TO PUMP
11 A
M E T A L TO GLASSSEAL
Figure 1.
Schematic diagram of system
C4Hl0. V e utilized the column described, operating a t room (25' C.) temperatures. Figure 2 s h o w the comparative chromatograms. Chroma6 ogram I \vas obtained with atmospheric output pressure and flame ionization detector, n-hile chromatograms 11-4 and B Tvere taken utilizing vacuum output and the QQ detector. The qualitative collector ( B ) of the QQ detector was tuned to 16 A l l U and was therefore selectively sensitive t o CH,. The higher speed of the analysis using the vacuum output chromatographic system is clearly indicated; the retention time values for the respective compounds are almost the half of those obtained when measuring with conventional system with atmospheric pressures. As seen on the chromatograms, the separating qualities using vacuum output are comparable to or better than those obtained when operating a t atmospheric output pressures. The calculated number of theoretical plates for the CH, peak was 19,000 using vacuum output and 15,500 a t atmospheric output pressure. The low theoretical plate numbers can probably be explained by the nature of the capillary column used. This column material is liquid only a t higher temperatures, and thus it is usually operated there. I n our measurements, however, room temperature was used to obtain base line separation for the C1 to C4hydrocarbons. We performed experiments to determine the effect of the input pressure on the performance of the vacuum output .system. In our measurement the input pressure was varied from subatmospheric up to 30 p s i g . , and the retention times and resolutions were compared. The experimental mixture contained normal CH,, C*Hs, and C,Hlo. The retention times observed, plotted against the input pressure values, indicated that the slope of the retention time curves did not change when the input was operated below atmospheric pressure. Chromatograms obtained a t different input pressures indicated that, operating the input of the column even a t subatmospheric pressure (620 torr), a good separation was still possible. These experiments indicated the feasibility of a chromatographic system wherein not only vacuum output but
I1 8
I,,, 7
MIH.
5
6
,/I .i
3
z
,
VbRT
~
TIHE
Figure 2. Separations achieved using conventional and vacuum output techniques
also subatmospheric, or perhaps also vacuum input, pressures are utilized. Modifications in the input part of our experimental system are necessary to explore the performance of a system wherein the input also operates in vacuum. The separation of permanent gases is performed very satisfactorily in packed columns. Their separation in capillary columns has not yet been studied, because the generally used ionization type detectors are not suitable for permanent gas detection; the
N2
question is, however, of great interest. The vacuum type QQ detector, which is able to detect any kind of gas, made this measurement possible. Utilizing the column described and operating a t room temperature (25" C.), identical retention times Kere found for H,, Sz, CO, and CH,. The retention time of COz, however, n as different and could be perfectly separated from any of the above mentioned gases. Figure 3 demonstrates the separation of S2 and COz a t room temperature (23" e.). The upper part of the figure sliow-a the quantitative Chromatogram obtained, while the lower part gives the curve obPRESSURE: tained by the qualitative collector Ivhich INPUT 3 0 p s i g was tuned to RI = 44 AIK, and was OUTPUT I ,103t0rr therefore selectirely sensitive only to
PO?.
2.
MIN. 3
2
I
Figure 3. Qualitative and quantitative separation of Nz and COZ a t room temperature
The sensitivity of this esgerimental vacuum output chromatographic system, utilizing a stream splitter, n a ~ . calculated and found t o be lo-" grams per second for Szand somen-hat higher for CH,. For a vacuum (input and output) chromatographic system calculations showed that the semitivity could be improved by two to three orders of magnitude. To achieve this, improvements in the sampling system and certain alterations in the pump cystem are necessary. I n our experiments the operation of the column output in vacuum was advantageous, maintaining separation and iensitivity as compared to measurements 17-ith atmospheric output pressure and decreasing time of analysis. More versatile detectors capable of measuring the quantities as well as the quality of both organic and inorganic compounds can be used. Another advantage is that this system enables operation of the VOL. 35, NO. 3, MARCH 1963
411
input of the column a t subatmoq~heric or vacuum pressures. This technique might introduce the pos4bility of extending the application of chromatographic compounds.
to
LITERATURE CITED
c.,
(1) Giddings, J. A ~ cHEhf. ~ 34, ~ ;311~ (1962). (2) Schalr, G., "Theoretische Grund1:Lyen
der Gaschromatographie," Deutsches Verlag der Wissenschaften, Berlin
iristable
(1960).
(3) V&radi, P. F., Ettre, X., A N ~ L . ,
CHEV.
34, 1117 (1962).
PETER F. Y ~ R A D I KITTY ETTRE
The Rlachlett Laboratories, Inc. Springdale, Conn.
An lodometric Method for the Macro- and Microdetermination of Peroxyd isulfate SIR: Methods reported in the literature for the determination of persulfates (peroxydisulfates) are based on the oyidation and titration of ferrous ion ( I ) , oxalate ( I ) , or iodide in strong acid ( 2 , 5 ) . Each of these, however, requires lengthy heating or reaction times, catalysts, and, for the ferrous or iodide methods, the exclusion of air. However, a t neutrality and with very high molar ratios of iodide to persulfate, iodide is oxidized quantitatively to triiodide without heat, catalysts, or interference by atmospheric oyygen. This reaction has been made the basis of a simple, rapid, and precise method which is applicable both to macro titrations, and to microspectrophotometry. Addition of 6Jf KI to a solution of persulfate, buffered a t p H 6.85, liberates 13-stoichiometrically. The liberated 13-is then titrated with SLO~-' or read spectrophotometrically a t 355 mp (3).
indicator may be used if deiiretl, but we have found it unnecessary. Micro Procedure. To 1.5 ml. of solution containing from 0.01 t o 0.15 pmole of persulfate add 1.O ml. of buffer and 0.5 ml. of 6 J I KI. Read abscxbance a t 355 mp from 5 to 30 minutes after addition of the K I . The color developed follon-s the BeerLambert law ( E = 28,600 liters mole-' em.-') in the Cary 11 (1-cm. cuvettes, 0.07-mm. slit width) and the Beckman DU (1-cm. cuvettes, 0.30-mm. slit width). With the Bausch and Lomb Spectronic 20-340 spectrophotometer (1.15-cm. cuvettes, 20-mp band width), deviation from strict linearity was noted above an absorbance of 0.4. hlso, probably as a result of its greater band width, the apparent absorptivity over the linear portion was somewhat, lower, 635, = 25,200 liters mole-' cm.-1 RESULTS
RELIABILITY O F h h C R 0 l \ I E r H O U . *\lthough the customary volumetric procedures were routinely employed with good precision (=k0.27, relati1 e standReagents. Potassium iodide soluard deviation), a gravimetric-spectrotion (iodate-free), 6 X . Sodium thiophotometric procedure was employed sulfate, standard solution, O.liYJ standfor increased precision in analyses comardized against re.;ublimed I,. Phosphate huffer, p H 6.85, O.lX, 0.05 moles paring the oxalate method. Solutions KHd'O, plus 0.05 moles K,HPO, per as well as solid reagents were weighed to liter. 4.; or seven significant figures on a Macro Procedure. Weigh accu$ingle-pan analytical balance. In this rately 1 to 2 minoles of (SI14i2S208, r a y the mas5 of thiosulfate solution KA08 or other persulfate into a 100-ml. equivalent to a known mass of resubflask and dissolve in 20 ml. of buffer. limed iodine was determined. -4dd 10 nil. of 6Jl KI and titrate xvith In titration, the unknown solution standard thiosulfate until the yellon rvas mighrd before and after titration color just disappear?. Soluble qtarrh EXPERIMENTAL
Table I.
Percentage Purity of Persulfates"
Snit
a
Oxalate method
96 48 i 0 97 27 rt 0 99 53 i 0 99 90 f 0 95 21 + 0 95 95 f 0 98 01 i 0 98 90 f 0 Standard deviations are given for the series
412
ANALYTICAL CHEMISTRY
Iodometric macro method
217 96 41 f 0 0 l C , 23 96 35 i 0 02 15 99 64 i 0 01 13 DO 97 i 0 01 22 95.19 + 0 02 17 95 99 f 0 01 20 98 05 =t0.02 11 98 97 rt 0 01 of five runs made on each sample.
and the mass, rather than the volume, of titrant added was obtained by difference. Titrations w r e carried only to the point of a very faint yellow color, rather than to complete loss of color to ensure that a ion- concentration of 1 8 remained. The concentration of the 13remaining was then determined spectrophotometrically a t 355 mp by comparison with known dilutions of standardized 11-solutions. I n this way it was possible to obtain replicate runs Ivith a precision of 10.02% or better. Because of 1o.s by evaporation with the hot oxalate solutions required, this same method was not applicable to the oxalate determinations. Results of five replicate runs on each of four samples of K2d208and four of (SHJLS20sdone in this way show excellent agreement (Table I), The samples were commercial mnples from four different sources. RELIABILITYOF XICHOMETHOD. Fresh persulfate solutioris, made from samples analyzed by the macro method, were determined by the micro method. The results obtained were both precise and accurate (relative error) to =tl% (relative standard deviation) n ith either the Cary or Beckrnan spectrophotomn-ith the Bausch and eters and to i~ta7~ Lonib spectrophotometer. The color de1 clops fully in leq5 than 2 minutes and is itablc for a t least 30 minutes. After that tinir a slow air oxidation of I- begins to liecome e l ident. PrmrIrsrnLE Y ~ R I ~ T I O SKeither ~. the macro nor micro methods were appreciably affccttd by changes of = t O 3 u1iit.i in the p H of the buffcr, k507c in its roncrntration or ilOyGin the concmtration of the K I . Thui. saturated solution< of KI (5,921a t 15" C., 6.111 a t 20" C., 6.2111 a t 25" C.) could he uwd rather than the 6Jf solution actually employed. SOLUTIOX STmILIrT. Solutions of K I autoxidize slon-ly on standing, cren if well stoppered in dark brown bottles. To obviate high blanks we discarded solutions when their absorbance a t 355 nip exceeded 0.1. This level of autoxidation, however, normally took many