A . E r i c P a r k i n s o n a n d E . C . W a g n e r
John Harrison L aboratory o f Chem istry, U niversity o f Pennsylvania, Philadelphia, P a.
T
H E bisulfite method of Ripper (27) was initially tested only with formaldehyde, acetaldehyde, benzaldehyde, and vanillin. A study b y Feinberg (7) included also salicylaldéhyde, p-hydroxy benzaldehyde, and anisaldehyde, results for these being only semi-quantitative. Jolies determined furfural and pentoses (IS) and acetone (12, 29) ; M eyer (25, 26) recommends the method for several aldehydes in addition to those mentioned, but without evidence that its applicability has been tested. The bisulfite method has been used for estimation o f lactic acid via acetaldehyde (2, S, 9 ,1 0 ), and o f some unsaturated aldehydes (11, 15, 26). Reports as to the accuracy of the method are conflicting, in some cases because of failure to appreciate the nature of the reactions involved and the conditions essential to accuracy.
Th e reaction between carbonyl compounds and bisulfite is reversible:
R .C O + -HSO, R .C (O H )S O
,-I I
The distribution at equilibrium varies with the identity of the carbonyl compound, the pH , temperature and concentra
tion o f the solution, and the excess of bisulfite. The results of analysis depend further upon the specific rates of the addi
tion and dissociation reactions, these also being affected by the conditions mentioned. Kerp and collaborators (14) made equilibrium studies of the bisulfite compounds of for
maldehyde (6), acetaldehyde, benzaldehyde, furfural, acetone, and glucose, and Stewart and Donnally (81) reported a more
elaborate study of benzaldehyde-bisulfite. Using K erp’s data, K olthoff (19) calculated the inherent error in analysis due to dissociation.
A consideration of available evidence permits the following
conclusions: '
1. The accuracy of analysts is determined primarily by the value of the equilibrium constant for the dissociation. When this is of the order of 10 ~4 or less, accurate analysis is feasible (formaldehyde, acetaldehyde, benzaldehyde, furfural). When K equals 10“ ’ (acetone), analysis is possible if bisulfite is used in large excess. When K is greater than 10~J (glucose), the results are too low.
2. The accuracy of the method is increased, especially when K is unfavorably high, by excess of bisulfite and by increase in the concentrations of both reactants. When K equals 10~T (formaldehyde) or 10 “ • (acetaldehyde), even dilute solutions can be analyzed using only a moderate excess of bisulfite.
3. The accuracy of analysis is affected by temperature, but here two conflicting effects are to be noted. The error due to dissociation can be decreased by working at low temperature, as the value of K decreases with decreasing temperature (14, 81), but the rate of the addition reaction is thereby decreased, and the time necessary for attainment of equilibrium may exceed the 15 to 60 minutes usually specified. The analysis of some alde
hydes requires a low temperature at the time of titration; in such cases the reaction liquid may be allowed to stand a suitable interval at room temperature and then chilled for a time be
fore and during titration.
4. The rates of addition and dissociation are affected by the hydrogen-ion concentration, the former decreasing with increas
ing acidity, while the latter and the dissociation constant are at a minimum somewhat in the acid region (at about pH = 1.8 for
Ta b l e I. An a l y s is o f Al d e h y d e s b y Mo d if ie d Rip p e r- Fe in b e r g a n d Ex c e s s- Io d in e Pr o c e d u r e s
benzaldehyde bisulfite), and increase as the acidity diminishes (14, SI, 8$). Practical application of these relationships was made by Tomoda (38) in the estimation of acetaldehyde, the excess bisulfite being titrated as usual, the pH adjusted to about 8 (sodium bicarbonate) to accelerate the dissociation of the bi
sulfite compound, and the bound bisulfite then titrated as a direct measure of the aldehyde. A similar procedure was described recently by le a (83) for heptaldehyde. Donnally (5) applied a more rigid control to the analysis of formaldehyde and benz
aldehyde, making three pH adjustments: one (sodium bicar
bonate) for the addition, a second at about pH 2 (acetic or phos
phoric acid) for titration of excess bisulfite, and a third (sodium carbonate) for titration of bound bisulfite. Such procedures are theoretically superior to the method of Ripper, afford a direct instead of an indirect analysis, and eliminate the blank analysis.
The accuracy, however, is not necessarily increased, for the direct titration starts at the end point of the excess-bisulfite ti
tration, and the imperfections in this end point will affect equally the results by both methods. Donnally obtained for formalde
hyde nearly identical results by both titrations, but results for benzaldehyde were high as calculated from the titration of excess bisulfite. This is attributed to oxidation of bisulfite by air, but it is not clear why such incidental oxidation did not similarly affect the results for formaldehyde.
5. An inherent source of inaccuracy in the Ripper method is the manner of titration— the addition of iodine to bisulfite—
when, as is well known, correct results are obtained only by the reverse procedure (17, S3). The irregularity has been assigned variously to reduction by the hydriodic acid of reaction, to loss of sulfur dioxide by volatilization, and to air oxidation of sulfite, the last being probably the chief source (S, 16, 80, 84). As al
addition of an aliquot of the reaction liquid to a measured excess of iodine and prompt back-titration with thiosulfate— was used by one of the writers (W.) in 1925 in the early trials of the excess- iodine method described below, and was independently adopted as preferable by Kolthoff (18) for analysis of formaldehyde, acetaldehyde, benzaldehyde, and furfural.
This paper presents results obtained b y the excess-iodine method ju st mentioned, in comparison with results given by a somewhat modified Ripper-Feinberg procedure, both applied to the following com pounds: formaldehyde, acetaldehyde, propionaldéhyde, n- and isobutyraldéhydes, n- and iso- valeraldehydes, n-heptaldehyde, acetal, benzaldehyde, salicyl
aldéhyde, vanillin, piperonal, paraldehyde, crotonaldehyde, and cinnamaldéhyde, the last three not being analyzable by the procedures used. The results show a marked superiority for the excess-iodine procedure. Under the conditions em
ployed, the acidity of the unbuffered bisulfite was somewhat too high for the m ost rapid attainment of equilibrium, and especially toward the end of the titration, probably exceeded the optimum for greatest stability of the bisulfite compound.
The unfavorable effect of the first condition was counter
balanced b y use of a rather large excess of bisulfite of greater concentration (0.3 to 0.4 M ) than was specified by Ripper or
T a b l e I ( Continued)
heating through column wall.
c Average of 7 determinations.
* Using the direct titration (bound biBulnte) method, Lea (28) obtained for heptaldehyde 19 results which ranged from 93.3 to 97.0 per cent and averaged 95.6 per cent.
I N ote irregular effect of double bond.
Feinberg. That of the second condition was decreased, in the analysis of aldehydes whose bisulfite compounds gave evidence of too rapid dissociation at room temperature, by chilling in ice (after standing at room temperature) for a short time before and during contact with iodine. (Analysis by Ripper’s method at low temperature has been recommended by several chemists, 4, 30, 34.) This precaution was ad
vantageous in the estimation of aromatic and higher ali
phatic aldehydes.
Rationalization o f the bisulfite procedure must await extension of methods such as those used by Kerp et ah, and b y Stewart and Donnally, to other compounds, so as to permit selection of optimum conditions for addition and titration.
This work is projected for the near future.
Ex p e r i m e n t a l
R e a g e n t s . In most of the trials the sodium bisulfite solu
tions were 0.3 or 0.4 M . They were not stabilized {SO, 16, 24), the effective concentration being determined for each series of analyses by means of a parallel blank. Sodium thiosulfate
Liquid aldehydes were Eastman Kodak Company products.
They were shaken with sodium carbonate solution, dried with cal
cium chloride, and specimens of good boiling point obtained by means of an 8-ball Snyder column. If necessary, specimens were redistilled, using a semi-micro column, until maximum results were obtained by analysis. Specimens which suffered rapid air-oxidation or polymerization were redistilled before each analysis and used promptly. Benzaldehyde was finally dis
tilled under nitrogen, and samples were collected directly in the measured bisulfite solution. The deterioration of most of these aldehydes was rapid, and in general it was found that analytical purity could be assumed only for specimens freshly purified.
The deterioration of several aldehydes was plainly due to oxida
tion. Some deteriorated samples left droplets insoluble or slowly soluble in bisulfite solution, suggesting the presence of polymers; the trials with paraldehyde indicated that aldehyde polymers cannot be estimated by the bisulfite method.
Vanillin and piperonal were Kahlbaum specimens, recrystal
lized from 10 and 50 per cent alcohol, respectively.
A n a l y t i c a l P r o c e d u r e s . Each analysis required two 50-cc.
brated against the flasks, and the titratable bisulfite determined as described below. In certain cases indicated in Table I the liquid was chilled in ice during the last 10 minutes of the standing period and throughout the determination of bisulfite.
About 2 cc. of formalin were introduced into a weighed 100-cc.
volumetric flask containing about 25 cc. of water, the increase in weight was determined and the solution diluted to the mark, and 10-cc. aliquots were used for analysis.
Freshly distilled acetaldehyde was transferred to a weighed glass ampoule, which was sealed off and reweighed. The ampoule was broken under bisulfite solution in the reaction flask, the solution diluted to the mark, and a volume of water equal to that displaced by the ampoule added from a Mohr pipet.
For water-soluble aldehydes the stoppered 50-cc. flask contain
ing 25 cc. of bisulfite solution was weighed, the sample dropped
436 A N A L Y T I C A L E D I T I O N Vol. 6, No. 6 into the liquid, and the flask reweighed. For aldehydes insolu
ble or slightly soluble in water, 5 cc. (in some cases 10 cc.) of aldehyde-free alcohol were stratified upon the bisulfite solution before introducing the sample. Solid aldehydes were intro
duced through a wide-stem funnel. When alcohol was used in an analysis an equal volume was added to the blank.
To determine excess bisulfite by the modified Ripper-Feinberg method, 10 cc. of solution were withdrawn, and the pipet was thrust nearly to the bottom of an Erlenmeyer flask and allowed to drain. Excess bisulfite was titrated at once and rather rapidly with 0.1 N iodine, with the addition of 5 cc. of 0.5 per cent starch indicator near the end point. With aliphatic aldehydes the end points were generally satisfactory. With aromatic aldehydes the end color was transitory, persisting only a few seconds;
unless such titrations were run rapidly the approximately cor
rect end point was inevitably overrun. To determine the effective strength of the bisulfite solution, an aliquot of the blank was similarly titrated.
T o determine free bisulfite by the excess-iodine method, 50 cc.
of 0.1 N iodine were pipetted into an Erlenmeyer flask, and a 10- similarly. The end points by this procedure were in all cases satis
factory.
Results of trials by the two methods are given in Table I.
The results in Table I show that the excess-iodine pro
cedure is in general superior to the modified Ripper-Feinberg method. The effect of time was not pronounced, and it appears that 45 to 60 minutes, and in some cases less time, will yield maximum values for most of the compounds tested.
The effect of low temperature was generally favorable though rather small for aliphatic aldehydes, but was pronounced for aromatic aldehydes, analysis o f which requires chilling of the liquid before titration. Neither method can be con
sidered precise, except for formaldehyde, and agreement of duplicates within 0.5 per cent is in general to be considered good.
De c o m p o s i t i o n o f Al d e h y d e Bi s u l f i t e s i n Co n t a c t w i t h Io d i n e
In the excess-iodine procedure some dissociation of alde
hyde bisulfite must occur between the moment the aliquot is added to excess iodine and that at which the end point of the thiosulfate titration is reached. In practice this is decreased by chilling the liquid. T o obtain an approximate idea as to the relative rates of dissociation of various aldehyde bisulfites in contact with iodine under the conditions of analysis, self-explanatory tests were made (Table II).
T a b l e II. D e c o m p o s i t i o n o f A l d e h y d e B i s u l f i t e s i n
The results in Table I I indicate that the bisulfite compounds of lower aliphatic aldehydes react somewhat slowly with iodine, formaldehyde bisulfite being wholly stable. The
bi-o r Ex c e s s Io d i n e® Co n t a c tw i t h Io d in e
sulfite compounds of aromatic aldehydes dissociate much more rapidly; in ice the rate is much decreased, but is still sufficiently high to show that even with prom pt titration of excess iodine a sensible negative error will enter the results.
Conclusions
The Ripper-Feinberg bisulfite method yields low results with many aldehydes because of the reversible dissociation of the aldehyde bisulfites, and because of inaccuracy in the titration of excess bisulfite b y iodine. T o avoid the titra
tion error a procedure is used in which an aliquot of the analy
sis liquid is added to a measured excess of iodine, and the excess titrated prom ptly with thiosulfate. With aldehydes whose bisulfite compounds dissociate too rapidly at room tem
perature the liquid is chilled in ice before and during contact with iodine. Both methods have been applied to a more ex
tended list of aldehydes than has previously been tested by the bisulfite m ethod; the excess-iodine procedure gave the higher and more consistent results and better end points.
The aldehydes tested, and averaged purity results by the excess-iodine method, are as follows: formaldehyde (100 per cent), acetaldehyde (98.9 per cent), propionaldéhyde (98.6 per cent), n-butyraldehyde (98.5 per cent), isobutyr
aldéhyde (95.8 per cent), n-valeraldehyde (96.3 per cent), isovaleraldehyde (97.6 per cent), n-heptaldehyde (97.2 per cent), acetal (97.2 per cent), benzaldehyde (97.1 per cent) salicylaldéhyde (96.1 per cent), vanillin (97.3 per cent), piperonal (100.6 per cent). Paraldehyde, crotonaldehyde, and cinnamaldéhyde could not be satisfactorily analyzed.
Literature Cited
(26) M eyer, H ., “ Analyse und Konstitutionserm ittlung organischer Verbindungen,” 5th cd., p. 451, J. Springer, Berlin, 1931.