improve the precision of analysis of samples of low oxygen content. DISCUSSION
This study, primarily aimed a t the reinvestigation of the Walter method, resulted in the successful elimination of the difficulties therein encountered. The troublesome use of graphite chips was completely avoided. Eliniination of the chips permitted better control of the contents of the crucible during analysis and facilitated degassing. Also, i t was shown that if graphite powder, used for thermal insulation, enters the crucible, satisfactory degassing becomes difficult. Up t o six samples were analyzed in a single run withno signs of incomplete extraction of gases. Reconditioning of the furnace between analyses was not required. A better understanding of the reaction conditions was gained. Figure 1 shows the bottom parts of the crucibles
TMinati on
subjected to different treatments. During degassing, the tin appears to have volatilized completely. The escaping vapor bombarded and conditioned the walls of the crucible. In the absence of tin flux, the titanium melted and spread but retained its metallic luster, showing practically no carbide formation. On the other hand, when tin flux was used, the sample residue appeared black and showed little or no metallic luster. Contrary to common belief, a sample fluxed with tin spread over less area than a sample without flux. The presence of tin mas evidently more essential than a large area of sample-graphite contact for completing the reaction. From this it was concluded that the graphite chips of the Walter method are superfluous, ACKNOWLEDGMENT
The authors are grateful to the many members of the Battelle staff who showed interest in the progress of the
work, and particularly to M. A. Van Camp and D. F. Kohler, whose cooperation contributed considerably to the success of this work. One of the authors (Ch. V.) gratefully acknowledges the fellowship granted by the International Cooperation Administration, Washington, D. C., and the deputation by the Atomic Energy Establishnient Trombay, Government of India, Bombay, India. LITERATURE CITED
(1) Albreoht, W. AI., hlallett, M. W., ANAL. CHEW26,401 (1954). ( 2 ) Hansen, W. R.,Mallett, 31.R., Ibid., 29,1868 (1957).
(3) Hanseii, W. R., Mallett, M. W., Traeciak, X I . J., Zbid., 31, 1237 (1959). ( 4 ) McDonald, R. S., Fagel J. E., Balia, E. W., Ibzd., 27, 1632 (1955). (5) Mallett, X I . Vi7., Griffith, C. B., Trans. Am. SOC.Metals 46, 375 (1954). (6) Walter, Dean I., ANAL. CHEM. 22, 297 (1950). RECEIVEDfor review June 6, 1960. Accepted Bugust 26, 1960.
of Forima Ide hyd by Gas
atography SIR: Schepartz and McDowell (6) recently reported an elution time for formaldehyde. The elution was obtained from a Carbowax 2Ohf (Union Carbide Chemicals Go.) column at 90" C. Kyryacos, Menapace, and Boord (5),in an earlier publication, presented chromatograms showing the separation of formaldehyde on an iso-octyi decyl phthalate column at 105' 6. but were not able to effect a quantitative estimation. In addition, Yokley and Ferguson (6) reported that some formaldehyde passes through a bis [2-(2methoxyethoxy)ethyl] ether column but were not able to resolve it a t 50' C. The authors investigated a variety of substrates from which formaldehyde can be eluted. These were not useful for quantitative work for reasons of relatively high substrate volatility-e.g., his [2-(2-methoxyethoxy)ethyl]ether a t 10.5" C. and Carbom-ax 2GM a t 120' C. -or poor resolution of formaldehyde in the presence of other polar compounds that occurs on a sorbitol or erythritol column a t 135' @. Recently we found that the surfactants present in several of the commercia! detergents can act as substrates from which formaldehyde may be suc1
ANALYTFCAL CHEMISTRY
cessfully eluted and estimated quantitatively. Desty and Harbourn ( 2 ) used the detergent Tide in granular form directly as a column packing to effect a variety of separations, not including that of formaldehyde. On the other hand, Decora and Dhneen ( 1 ) used Tide support only, coated with other substrates, for the separation of pyridines, I n the present investigation, the surfactant \vas extracted from the detergent pom-der and coated on such standard supports as C 22 firebrick hnd Fluoropak 80. This was necessary to reduce substantially the tailing of the formaldehyde and vater peaks which was always obtained on the heattreated detergent powder packed columns, Data obtained from two different columns are given below. One column was used to show that formaldehyde could be separated and quantitatively determined from its solution in v-ater. The other was operated to demonstrate the qualitative analysis of some complex solutions containing f crmzldehyde. ~ X P ~ ~ I M ~ N ~ A ~
The detergent powder wm heated, first at 125' C.and then at 180" C.,
for 24-hour periods to remove volatiles. The surfactant present nas subsequently extracted with petroleum ether or benzene. I n the quantitative work a borosilicate glass column 4 feet long and with a 6/32-inch bore was used. This was packed with a mixture of 30to 60-mesh Fluoropak 80 coated with 10% by weight of Tide surfactant and 30- to 60-mesh sand coated with 1% by weight oE the same surfactant, in the proportion, two volumes of the former to one of the latter. The coated sand was required to prevent compaction of the Fluoropak with its resulting nonuniformity of packing and unreasonably high backpressures. For the qualitative analyses, a copper column 15 feet long and with a bore of l/4 inch was packed with 30- to 60mesh C 22 firebrick coated with 30% by weight of the surfactant obtained from Sail detergent powder (A & P Food Stores). The active material is a blend of sodium tridecyl- and dodecylbenzene sulfonates, manufactured by Ultra Chemical Works, Inc. Columns containing such a blend gave separations which were as good as those given by the extracted material. Hence, heating the detergent apparently does not substantially affect, the composition of the surfactants. Preliminary work indicates that the surfactants obtained from such commer-
9
4
2 L W K
Z
0
c 0
1
2
TIME, MIN.
Figure 1 . Representative chromatogram Quantitative
analysis
of a 20.15% formaldehyde-water solu-
tion
cia1 detergents as Tide, Sail, and Fab on either support are similarly effective both for the qualitative and quantitative analysis of formaldehyde-containing solutions. The instrument used in these studies was a standard Model I< 1 Kromotog (Burrell Corp.). RESULTS
Quantitative Study. Analyses were run on each of five solutions of formaldehyde in water ranging in concentration from about 1 to 20% b y weight, which were made by dilution of a 20.15% stock solution. T h e stock solution was prepared by boiling paraformaldehyde in distilled water and filtering, its formaldehyde content being determined b y the standard hydroxylamine niethod. A typical chromatogram is given in Figure 1. This was obtained with helium flowing a t 20 ml. per minute
(exit conditions) (He backpressure 3 p.s.i.) and the column a t 145' C. The sample volume was 2 pl. These conditions were used to allow the water, which tails considerably, to pass through the column in the shortest time consistent with good separation. The elution time for formaldehyde in these conditions is 25 seconds. The corresponding elution time for air is about 4 seconds. The very small air peak is not apparent in Figure 1, because of the decreased recorder sensitivity required for the analysis shown. The formaldehyde peak was identified b y applying the specific reaction with chromotropic acid to a sample trapped out of the column effluent. The response to formaldehyde as determined by this method was linear in the range 0.01 to 0.5 mg. of formaldehyde. At least 10 samples of each solution were analyzed with a coefficient of variation in the results of not more than =t3% and usually less than +1.50100. Qualitative Analyses. The Sail surfactant supported on C 22 firebrick produced a general-purpose column for t h e separation of many polar and a few nonpolar compounds. Thus, i t has resolved mixtures of t h e aliphatic aldehydes through isovaleraldehyde, several substituted pyridines, and a mixture of tertbutyl hydroperoxide, di-tert-butyl peroxide, and tert-butyl alcohol, and shows promise in t h e analysis of coolflame combustion products. In Figure 2 an example is given of the qualitative analysis on this column a t 95' C. of a synthetic blend of pen-
le Systems in SIR: Previous theoretical calculations (1, 8 ) of currents for voltammetry with linearly changing potential with electrochemica~ly irreversible reactions have obscured through their mathematical formulations certain features of the potential-time relations to be expected in such cases. Because these features are useful in making qualitative distinction between reversible and irreversible systems and further allow the simple determination of kinetic parameters in irreversible cases, we consider them here. Statements of the diffusion problem with boundary conditions have been given for planar electrod~sby Delahay (1) and for spherical electrodes by DeMars and Shain ( 2 ) . One of these
1.0 r
TIME, MIN.
Figure 2. Chromatogram of a synthetic blend containing formaldehyde tane, formaldehyde, acetone, methanol, ethanol, and water. At this temperature water is retained for an additional period of about 25 minutes. Hence, although i t tails badly, its presence in the mixture does not interfere with the analysis for the other components. Programmed temperature operation is indicated for the analysis of more complex mixtures containing water. LITERATURE CITED
(1) Decora, A. W., Dinneen, G. U., ANAL. CHEM.32,164 (1960).
(2) Desty, D. H., Harbourn, C. L. A., Ibid., 31, 1965 (1959). (3) Kvryacoe, G., Menapace, K. R., Bodrd, C. E., Ibid., 31, 222 (1959). (4) Schepartz, A. I., RIcDowell, P. E., Ibzd., 32,723 (1960). (5) Yokley, C. R., Ferguson R. E., Combustion and Flame 2 , 117 (1958). QANDLER SAMUEL ROBERT STROM Departments of Chemical and Mechanical Engineering University of Toronto Toronto, Ontario, Canada RECEIVEDfor review August 29, 1960. Accepted September 29, 1960.
tationary Electrode
conditions relates the flux of the reactant at the surface to the rate of the charge-transfer process. By assuming rate constants of the form predicted by absolute rate theory, this condition can be stated as follows: t>O,r = 0 D(bC/bz)
=
k,C exp(pt)
(1)
where the symbology is that of Delahay (1). D is the diffusion coefficient of the reactive species; C, its concentration; X, linear distance; t, time; ki, the rate constant a t the initial potential; p, the product of d E / d t and d In k/dE, where both differentials are assumed constant; and E, the applied potential. If the Jurface concentration were
maintained a t its initial value throughout the electrolysis, Equation 1 predicts that the current would be an exponentially increasing function of potential, but would be independent of scan rate and diffusion coefficient. I n practice the passage of current causes the surface concentration to decrease with time and the experimental current passes through a maximum. The amount by which the concentration is decreased from its initia! value a t any specific potential is related roughly to the total amount of reaction prior to the time a t which that potential is reached. The deviation is therefore inversely related to the scan rate because the maximum instantaneous current a t any potential VOL. 32,
NO. 13,
DECEMBER I960
1893
