11). When 10 ml. of diphenylamine reagent is heated with 2 ml. of 40 y per ml. methylcellulose solution, the absorbance is 357, higher than' that obtained Kith 5 ml. of reagent; the absorbance per microgram per milliliter of final solution increases 2.3fold. The use of 15 ml. of reagent does not further increase the absorbance per microgram per milliliter of the original solution. The combination of 2 ml. of methylccllulose and 5 ml. of diphenylamine reagent heated for 30 minutes a t 108.3' C. was selected because of the convenient volumes and concentrations in the experiments where the method mas applied (3, 4). The reproducibility among samples of the same concentration run as separate sets with different combinations of reactants, time of heating, and temperature mas evaluated and compared by the F test with the reproducibility of 2 ml. of methylcellulose heated with 5 ml. of diphenylamine reagent for 30 minutes a t 108.3' C. (Table 11). Varying volumes of reactants, heating time, and temperature does not alter the reproducibility among multiple samples, except for the use of 1 ml. of methylcellulose and 6 ml. of reagent. Here the variability among samples of the same concentration is greater than 2 ml. of methylcellulose and 5 nil. of reagent (Frat,,> F z q 0 ) . With temperature constant (108.S0C.),
the absorbance increases rapidly during the first hour of heating; a t 30 minutes, the absorbance is increasing a t the rate of 3.8% per minute (Figure 2). Further heating produces relatively small increases in absorbance only after 2 hours. Reproducibility among multiple samples of the same concentration was not improved by a 1- or 2-hour heating us. a 30-minute heating (Table 11). With time constant (30 minutes), there is a curvilinear increase in absorbance with bath temperature (Figure 3). The absorbance per microgram increases tenfold between 90' and 110' C.. and a t 108" C. there is a 8.7% increase in absorbance per degree rise in temperature. The temperature of the bath, as such, is not crucial, because a set of standards is included in each heating, but for reproducibility within a given heating, all tubes must be exposed to the same conditions for 30 minutes. These conditions are established by adequate stirring and by maintenance of uniform bath temperature within 0.1' C. The precision of the method is indicated by the average deviation of the absorbance per microgram as a per cent of the mean (calculated as in Table I). The per cent average deviations for 158 sets of standards Ivere summated and the mean of the series was +2.1%.
LITERATURE CITED
(1) Alving, A. S., Rubin, J., Miller, B. F., J . Biol. Chem. 127, 609 (1939). (2) Berger, E. Y., unpublished observations. (3) Berger, E. Y., Kanzaki, G., Homer, M. A., Steele, J. M., Am. J . Physiol. 196, 74 11959). (4) Berger,' E. Y., Steele, J. Id.,J . Gen. Physiol. 41, 1135 (1958). (5) Bridges, R. R., ANAL.CHEW24, 2004 (1952). (6) Dische, Z., Mikrochemie 7, 33 (1929).
( 7 ) Dreywood, Roman, ISD. ENG.CHEM., ASAL. ED. 18. 499 11946). (8) Harrison, €1: E., Proc. doc. Erptl. Biol. M e d . 49, 111 (1942). (9) Koehler, L. H., ASAL. CHEM.24, 1576 (1952). (10) Loewus, F. A., Zbid., 24, 219 (1952). (11) McCready, R. M., Guggolz, Jack, Silvera. Vernon. On-ens. H. S.. Zbid.., 22., 1156 (1950). ' (12) hlorse, E. E., Zbid., 19, 1012 (1947). (13) Sanisel, E. P., DeLap, R. A,, Ibad., 23, 1795 (1951). (14) Scott, T. A., Jr., Melvin, E. H., Ibid., 25, 1656 (1953). (15) Seifter, S., Dayton, S., Kovic, B., Muntwyler, E., Arch. Biochem. 2 5 , 191 (1950). (16) Shetlar, M. R., ANAL. CHEM..24, 1844 (1952). (17) Somogyi, M,, J . Biol. Chem. 86, 655 (1930).
RECEIVED for review January 16, 1959. Accepted March 31, 1959. From the Department of Medicine, Sew York University College of Medicine. Research supported in part by a grant (A-311) from the Kational Institute of Arthritis and Metabolic Diseases, Public Health Service.
Stoichiometry of Chlorite-Aldehyde Reactions Analytical Procedures HERBERT
F. LAUNER and
YOSHIO TOMIMATSU
Western Regional Research Laborafory, Albany 7 0, Calif.
b Sodium chlorite showed no fixed stoichiometric relationship to aldehyde groups in reactions with aldoses, benzaldehyde, and dextran dialdehyde, not because of overoxidation but presumably because the unstable chlorine intermediates react with chlorite to varying extents. Stoichiometric ratios varied with aldehyde, kind and concentration of buffer, reaction rate, and other factors. Phosphate buffer 0.5M yielded ratios of 2.63 to 3.77, limiting its usefulness to known aldehydes or to those whose oxidation rates resemble those of known aldehydes. Phosphate buffer (3M) yielded ratios ranging from 2.56 to 2.79, permitting the use of a mean value in the analysis of unknown aldehydes including polysaccharides, with an uncertainty near
=t2.6% a t 0.06 mM aldehyde. Analytical procedures, determinations of ratio-time curves, and residual aldehyde concentrations are discussed.
I
attempt to develop an acidic micromethod for the determination of aldehyde groups in alkali-sensitive polymeric carbohydrates, glucose was initially used as a model substance. The results of studies of the reaction between phosphate-buff ered sodium chlorite (h'aCIOz) and glucose, from the practical analytical (5) and fundamental kinetic (6) standpoints have been described. Chlorite was adaptable to a precision of a few per cent a t 5 y of glucose per ml. with a corresponding precision to 0.6 y per ml. K AK
Using acetate-buffered chlorite a t a higher concentration, Launer, n'ilson, and Flynn ( 7 ) studied various procedures for determining glucose and, to a limited extent, cellobiose, melibiose, maltose, and lactose, and developed both photometric and volumetric methods. Stitt, Friedlander, Lewis, and Young (10) used a photometric procedure for applying chlorite to the determination of glucose in the presence of an excess of a ketose and fructose, and discussed the theoretical and quantitative aspects of the reactions involved. I n no study was overoxidation of the aldehyde group past the carboxyl stage observed. This conclusion appears warranted, although reagent decomposition during and after oxidation complicated the results. It is also in VOL. 31, NO. 8, AUGUST 1959
1385
3
-
0
z2 0.5M0uffer pH2.4 50'C. C. 0 . 0 0 0 0 S o d i u m Chlorite A. 0.0001 M Aldehyde
0.5M Buffer p H 3.1 25'C. C . 0.01 M Sodium Chlorite A. 0.0005 M Aldehyde I
2
I
l a Benzaldehyde 2 a Arabinose
Benzaldehyde Arabinose
Xylose Mannose 5 Rhamnose 6 Glucose 7 Lactose E Maltose
3 4
IO
eo
30
40
5 0
60
HOURS HOURS
Chlorite-aldehyde ratios for various aldeFigure 1. hydes under OSM, 2 5 " C. conditions R, values. 3.77,3.69, 3.56, 3.51, 3.36, 3.02,3.01, and 2.63 for Experiments 1 to 8, respectively
agreement with the results of this investigation for simple aldoses, an aromatic aldehyde, and a highly complex polyaldehyde under conditions which permitted no appreciable reagent decomposition, and is strongly indicated for a series of periodate oxycelluloses studied by Davidson and Neve11 ( 2 ) . Furthermore, experiments with lactones of the aldonic acids of rhamnose, mannose, lactose, melibiose, and glucose showed no Oxidation of the carboxyl (7). Chlorite is a unique reagent for aldehyde groups, because it is very sensitive, apparently highly specific, and functions in a weakly acidic medium which avoids the side reactions often undergone by aldehydic polysaccharides in the usually alkaline media of other reagents. Its stoichiometry, however, has not been clarified. Extensive work in phosphate (6, 6) and acetate (7) buffers gave a stoichiometric ratio within a few per cent of 3 moles of chlorite per mole of glucose. A few experiments ( 7 ) indicated a ratio of 3 for cellobiose, maltose,'and lactose. However, results (8, IS) with arabinose, xylose, galactose, mannose, rhamnose, and glucoheptose showed that the ratio 3 was exceeded, as do the recent results of Wilson and Padgett (14). This report shows that the stoichiometric ratios, R, for a group of aldehydes (including aldoses and dextran 1386
ANALYTICAL CHEMISTRY
Figure 2. Chlorite-aldehyde ratios hydes under OSM, 50" C. conditions
for various alde-
R,values. 3.50, 3.65, 3.30, and, 3.06 for Experiments la, 20, 5 a , and 60, respectively
dialdehyde) vary with the oxidation rate of each in 0 . 5 N phosphate buffer, and that variations are much smaller a t higher buffer concentrations. Other factors affect the ratios also, some in a n apparently complex manner. However, the principal emphasis is on rate studies of R values of simple substances under three sets of experimental conditions. These conditions appear applicable to the analysis of similar simple substances, and application to polysaccharides is intended for a future report. MATERIALS
The benzaldehyde, D-galactose, Dmannose, L-rhamnose, L-arabinose, and D-maltose (the last two were recrystallized once) were Eastman Kodak Co. White Label grade. The lactose was Mallinckrodt's analytical reagent grade (Mallinckrodt Chemical Works). The glucose was an NBS standard sample. The D-xylose was a commercial product, highly purified and tested. The dextran dialdehyde was a high molecular weight product of the reaction between sodium metaperiodate and the microbiological derivative of sucrose, dextran (4, 9 ) , with nominally 2 aldehyde groups per anhydroglucose unit. This extremely alkali-sensitive material was synthesized at the Northern Utilization Research and Development Division of u. S. Department of Agriculture. The solution was made up
with mild heating to contain 51.7 mg. (moisture-free basis) per liter and used in 3.1- and 10-fold dilutions t o give 0.0002M and 0.00007M aldehyde groups. All aldehyde solutions were let stand at least 1 day and were discarded upon indication of aging. Two concentrations of buffer, 0.5 and 3 M , representing the molarity of sodium dihydrogen phosphate in the reacting systems, were used. Various p H levels were obtained by varying the concentration of the other component, orthophosphoric acid. Preparation of the 0.5M buffers has been described (6). The 3 M buffer was identically prepared and purified except that a stock solution consisting of 82s grams of sodium dihydrogen phosphate monohydrate and 19.6 grams of 857, orthophosphoric acid was made up, which, upon 1 to 2 dilution became 3.1.1 with a p H of 3.20 a t 50' C. (3.14 a t 25' C.). The sodium chlorite (Mathieson Alkali Works "analytical" grade) was found to be close to labeled purity, and its solutions and those of the aldehydrs were made up 1 to 1.5y0stronger than the nominal values given to one significant figure in the drawings. EXPERIMENTAL PROCEDURE A N D TREATMENT OF DATA
Rate studies at 25' C. and 503 ==I 0.02' C. consisted of measuring the chlorite remaining a t intervals. The various starting conditions are given in Figures 1 to 4. After the reaction was stopped, either by chilling the
3
0
5.2 K
2il"l /
05M Buffer
Expt 9
IO
II
1
I t 13 14
1,"
Aldehyde
:
Dextran Olaldehydc
pH
Tamp. 'C.
3,.:
25
,I
I,
II
,,
I,
I,
4.0
Arabinose
8,
50
0.8
0;2
I .5
::
10.0 8
I,
"
2.4
I'
17 18 19
AO
mllllrnolar
3.0 5.0
" 'I
C.
"
0.0040 ,0024 ,001 6
20
,0008
"
07 ,061 ,051
30
60
90
I20
150
I80
HOURS
Figure 4. Chlorite-benzaldehyde buffer at various values of Co
R, valuer. 3.27,3.08, 2.89, and 2.55 for Experiments 17 to 20 respectively
HOURS
Figure 3. Chlorite-aldehyde ratios in 0.5M buffer under various conditions R, volues. 3.77,3.62,3.56,3.42,3.21, 3.02, 3.00,and 2.91 for Experiments 16, 13, 12, 1 1, 10, 9, 14, and 15, respectively
50" C. diquots (all samples were 20 ml.) with ice, or by raising the p H of the 25' C. aliquots to 5 with 0.15 gram of sodium bicarbonate, the by-product gaseous C102 was removed with a stream of nitrogen and the chlorite v a s determined by adding potassium iodide, nhich had been acidified immediately before use, and titrating the liberated iodine with 0.005, 0.01, or 0.02N, thiosulfate depending upon the initial chlorite concentration. Further experimental details were the same as those described for glucose (6), except that the experiments a t higher buffer and chlorite concentrations required more acidified potassium iodide: 1.6 grams in 15 ml. of 6N sulfuric acid for the 31M buffer, and 2.4 and 0.8 grams in 20 ml. of 2.4N sulfuric acid for the 0.5M buffer with initial chlorite concentrations of 0.01111 and 0.0008M, rcspectivel y For each point on the rate curves a test determination (with aldehyde) and a control determination (without) n-ere made as above, as was a determination of initial chIorite for each experiment. This gave results in milliliters of thiosulfate: VT, V,, and Po nil.. respectively, corresponding to Cr, C,, and Co chlorite concentration in moles per liter. The control is necessary because the reagent decomposes, although chlorite is satisfactorily stable (and unreactive) until acidified by the buffer. Thus the chlorite disappearing in the tcst must be corrected for de-
ratios in 0.5M
composition to yield CA, the chlorite disappearing because of aldehyde. It has been shown (5, '7) that C A can be calculated despite simultaneous reagent decomposition, using Equation l, with the symbols just described:
Under certain experimental conditions, reagent decomposition was negligible, and then C. = CO,and Ca was equal to the changes in CT with time. C A plotted against time would be the rate a t which chlorite is consumed by aldehyde (and by unstable intermediates arising therefrom), but corrected for reagent decomposition. However, a more useful function, the ratio, R = Ca/Ao, where A. is the initial aldehyde concentration, was plotted. This ratio, in terms of milliliters, V , of thiosulfate of normality iV for 20ml. aliquots is given in Equation 2, Ratio, R =
where the factor 4 is the relation between equivalents of thiosulfate and moles of chlorite. This equation embodies Equation 1 after transforming to accommodate reciprocal volumes. These facilitate plotting, because straight
lines are obtained for the decomposition phase, because of its second-order nature (7). Figure 5 shows the plots of the reciprocals of raw titration data for arabinose. Experiment 22, which was typical. The deviation of experimental points from the curves of Figure 5 affords some idea of typical precisions From these smooth curves the ~ / V T and l/TYcvalues n'ere taken and substituted to Equation 2 , resulting in curve 22, Figure 6. I n this manner all ratio-time curves were obtained. I n the experiments of Figure 1, the differences between V , and VT ranged from 4.5 to i . 5 ml. and for those of Figure 4, the differences were near 2 nil., each with a precision of approximately 0.5%. A plot of R against time represents the reaction rate divided by a constant, Ao, and thus a series of such curves involving the same A. represents the relative reaction rate of the series. However, for experiments with differing Ao, and for obtaining absolute rates, the relation CA = R A. niust be used. The value of R at the maximum of a curve will be referred to as R , and as such has a special significance in the discussion of results. I n a few experiments the reaction was studied by measuring the decrease in total oxidizing titer. The oxidation states above chloride, C1--namely, chlorate, chlorine dioxide, chlorite and, if any, hypochlorite and chlorinewere determined as an unresolved group by the addition of ferrous iron VOL. 31,
NO. 8, AUGUST 1959
1387
Figure 6. Chloritealdehyde ratios for various aldehydes in 3M buffer R,
2.79, 2.75, 2.76, 2.76, 2.74, 2.79, 2.66, 2.66, and 2.56 for Experiments 21 to 29, revalues.
0
;2
a
5
IO
I5
50'C.
C. O.OO2M Sodium Chlorite A. 0.0001 M Aldehyde
I
06
p H 3.2
3 M Buffer
spectively
2I 22
Benzaldehyde Arablnose
23
xylose Golactose Mannose Rhomnose Glucose
24 25 26 27 28 29
20
Lactose Maltose
HOURS
Figure 5. Rate of disappearance of chlorite, in terms of reciprocal milliliters of 0.0 1N thiosulfate, from a typical system (Experiment 22) 0.002M chlorite, pH 3.2,50' C., 3M phosphate buffer, 0.0001M arabinose in test (no aldehyde in control)
and back-titmtion with permanganate, using a procedure recently developed (1%'). Because reagent decomposition does not affect the total oxidizing titer, control experiments were not necpssary. CHANGE IN RATIO WITH CHANGE IN RATE IN 0.5M BUFFER
The reaction rates vary greatly n i t h the nature of the aldehyde, as shown in Figure 1, in which the rates for the two extremes, benzaldehyde and maltose, differ more than 50-fold. The conditions of Figure 1 are referred to as 0.531, 25" C. The ratio values a t the maxima, R,, in Figure 1 Ivhich correspond fairly closely to the number of moles of chlorite reacted per mole of each aldehyde, increase with the rate of oxidation of the aldehyde, regardless of vihether it is a simple aldehyde, a monosaccharide, a disaccharide, or, as will be shown later, a highly reducing polysaccharide. Essentially the same rate-ratio order and similar R,, values mere found for the much lower chlorite-aldehyde concentrations of Figure 2, which condition is referred to as 0.5M, 50" C. For complete agreement the R , value for benzaldehyde is 6% low, which may be a real effect or may be due to experimental difficulties a t relatively high reaction rates. The correlation between rate. as determined by the nature of the aldehyde. and ratio suggests that the ratios for two or more aldehydes will be the same if the rates of oxidation are the same under 0.5M conditions. Wilson and Padgett ( I C ) , using a 0.4M acetate buffer (Y), noted no such rateratio relationship. They concluded that 1388
ANALYTICAL CHEMISTRY
3
6
12
9
IS
18
I
I
HOURS
arabinose. xylose, mannose. rhamnose, and other fast sugars, reacted per mole, with 3.62 moles of chlorite, and that they as a group react in a mechanism with chlorite that differs from that of glucose with a ratio of 3.0. The effect is not simply one of rate. I n Experiment 9, Figure 3, chlorite was reduced over three times as rapidly as in Experiment 14, yet the R , values were practically identical. Increasing reaction rates five- to 10-fold by similarly increasing the concentrations of arabinose. xylose, glucose, and dextran dialdehyde, did not affect R,. I n an extreme case, d i e n arabinose vias increased 30-fold, R, actually decreased from 3.5 t o 3.2, although the major part of this effect was due t o the lower chlorite concentration in the more concentrated arabinose system. Because of experimental limitations, the chlorite concentration had decreased 35% in the higher arabinose system compared t o 1.5% in the lower, before two thirds of the aldose had been oxidized in either case, although Co was the same. The same effect of decreased chlorite was probably reflected in the results of previous workers (14) who reported a considerable decrease in ratio for an eightfold increase in xylose concrntration in acetate buffer. I n any case, the effect of varying aldehyde concentration over wide limits does not affect R, values in phosphate buffer systems, other things being equal. The rate-ratio relationship is complex. I n the case of fast aldehydes, periodate dextran, benzaldehyde, and arabinose (Figures 3 and 4) , R, n as considerably increased by increasing chlorite (Experiments 9 t o 13 and 17 to 20) and acidity (Experiments 15 and le), both of which increase the rate by increasing the concentration of chlorous acid, HC102, the principal oxidizing species. I n contrast, R, for slow aldehydes is either
insensitive to rate or actually shows a reversal of effect. Previous work with glucose (5) a t 50" C., shows that, despite large changes in chlorite (6.4-fold), glucose (35-fold), and hydrogen ion concentration (10-fold), R, remained 3.0. On the other hand, a t 25' C., in a series of experiments similar to Experiment 6 (Figure l ) , R , was decreased to 2.87 when Co &-as doubled, and decreased to 2.70 when the pH was changed to 2.4. Maltose also showed a decrease in R,, when Co was increased.
CONSTANCY
OF
RATIOS
IN 3M BUFFER
R , differences are considerably minimized in 3111 phosphate buffer, as seen from Experiments 21 t o 29, Figure 6, which shows an R, range of 2.56 to 2.79, compared to that in 0.5M buffer (Figure 1) of 2.63 to 3.77. This decrease in range is largely due to large decreases in R, of the faster aldehydes. All R, values were lowered by the change from 0.541 t o 3M buffer. T'ariation in A. but not Co or pH was studied a t 3M. Experiment 22 was repeated but the initial arabinose concentrations TTere twice, one half, and one fourth. S o effect upon R, was observed over this eightfold range. A11 such effects ~vouldbe expected to be smaller than a t 0.5-U. A variety of phosphate buffer concentrations and other buffers were studied. With other factors the same as those of Figure 6, sulfate-bisulfate buffer, a t 0.5 and 2M yielded chlorite-aldehyde ratios as high as 5. Phosphate buffer a t 4.4-11 and 50" C. yielded ratios of 2.50 for benzaldehyde, 2.46 for arabinose, and 2.55 for glucose, but the recession portions of the curves (to be discussed) differed more than a t 3M. Other systems, a t 40" C., pH 2.8;
.io" C., p H 2,3; and l.5M buffer a t 40" C., pH 2.4, were less desirable. .4t S M buffer (highly supersat,urated) (dextran dialdehyde yielded a ratio of 3.15. In 0.1M phosphate buffer the rate of decortiposition of chlorite is very high, especially when stray light has access to the y.stem ( 5 ) . RECESSION
OF RATIOS WITH TIME
.inother variability of the ratioswcession after the maximum-occurred in all experinierits in which simultaneous iwgent decomposition was measurable. Experiments 9, 10. 14, 15, 18, 19, and '70, in which reagent' decomposition, hecause of experimental conditions, was negligible over loug periods, reached constant ratios after the aldehyde had practically disappeared. Experiment 17 showed a slight recession because its higher Cocaused R slight but measurable decomposition, Ot5yc in 50 hours. These experinients, prolonged many hours after thr tinie required for cornplete oxidation, lleriioiistrate the extreme ,specificity of the reagent' for the aldehyde groups ~f wbstances of varying characteristics itnd complexities. I n all other experiment's recession was expected, and reflects the fact that test and control ,iecoinpose according to the second-ordtrr rate law. This law requires that C r rind C, tend t'o coni.erge, after reduction bj- aldehyde becomes slow lconiparcd to the rate of decomposition. 'l'liis may be expressed quantitat'ively 1)- applying the secondordpr rate law to Equation 1, giving
n-here R1 is the ratio t hours after R,, the maximum, ko is the decomposition rate constant, and h, the acidity function of the deconiposition, = (H+)/ [ K ~ O L4-O ~(")I (6). Values used for R H c ~ o ~ the , iomzation constant for chlorous acid, were 0,019 a t 25" C. and 0.008 at 50" C., approximated from the data of Barnett (1'1 for high salt concentrations. This gave h l . 5 = 0.023, 0.19. and O.OOS1 for Experiments 21, 16, and 4. respectively, whose k o values n-ere taken as 400, 320, and 55, in the same order Values calculated for R , from Equation 3 show no systematic disagreement with experimental ones. Typical niaximum deviations were 0.5, 2, -5, 5 -6, and -7% for Experiments 16, 4, 21, 2, 22, and 23, respectively. I n those experiments in which reagent decomposition was negligible, the product, t k~ CT hl.5, n-as very small compared to unity and R t = R, for all ohserved values of t. COMPLETENESS
OF
OXIDATION AT MAXIMA
Davidson Rnd Sevell ( 2 ) reported
incomplete oxidation of hydrocelluloses and periodate celluloses a t very high chlorite concentrations. Data are at hand for assessing the completeness of oxidation a t the maxima. By differentiating Equation 1 with respect to time and combining with the differential equation expressing the rate of oxidation of an aldehyde by chlorite ( 6 ) Equation 4 for the concentration of residual aldehyde a t the maximum i~ obtained:
arabinose (Experiment 15) A,,/Bo was 0.3%, whereas for Experiment 16, which n-as the same except that the pH m-as 2.4 instead of 4.0, and therefore much faster, rl,/Ao was over 10 times as high. Eightfold variation in arabinose concentration in experiments like 22 made no difference in A,/& I n the present connection residual aldehyde is of theoretical interest only, because the R factors are determined with known amounts of aldehyde.
(4
ATTEMPTS AT CHLORITE METHOD UNAFFECTED BY VARIABLE R,
The value fur k D ~ R giieii Q ahoie and koz i. the cwdation rate conqtant whose values for benzaldehyde, arabinose. glucobe, and maltobe nere approximately 53,000. 29.000 3500, and 2800, respectively for Experiments 21. 22, 27, and 29 a11 rate constants having the dimensions mole-l liter' hour-' h 0 ' b = 0.15. BJ substituting into Equation 4 the chlorite concentrations a t the maxima A,/& values were calculated to be 0.3, 0.5 3.2, and 5.3%, respectively. for Experiments 21, 22, 27, and 29 For the same substances a t 0.5M. Experiments 1, 2, 6, and 8, the residual aldehyde a t R , \\-a. practically the same, except that benzaldehyde was much lower-0.04, 0.46, 3.5, and 5.iSo,respectively. The amount of residual aldehyde a t R , is nearly proportional to the reciprocal of the rate of oxidation. I n accord with the first-order kinetics of aldehyde oxidation, allowing the reaction to proceed again as long after the maximum would lower A,/Ao to approximatel? the square of the values a t maximum, or 0.0009, 0.0025, 0.1, and 0.3Y6, respectively. This effect does not explain the results of Davidson and Neve11 (d), and in those cases it appears possible that the persisting residual reducing power after chlorite treatment, as measured by the unspecific copper tartrate reagent, was nonaldehydic and therefore not subject to oxidation by chlorite. To the extent the aldehyde is still present a t the maxima, the masinium values differ from the true ratios of Ca:Co. Correcting the glucose and maltose maxima (Figure 6) for A,, the R, values become 2.75 and 2.70, respectively. cloqe t o those of the other aldehydes. The oxidation of residual aldehyde after the maximum tends t o decrease the rate of curve recession. This effect, hon-ever, is very small, amounting a t the most to only some 0.04 unit per 3 hours for maltose. For those experiments in which reagent decomposition and, therefore, curve recession were negligible, A , values were small. For benzaldehyde and dextran dialdehyde, A,/& was calculated t o be 0 0 1 to 0.047'. For
It would be highly desirable to employ chlorite in a procedure unaffected by variable R,. Because all ratios greatly exceeded 1 to 1, it is obvious that chlorite, aside from oxidizing aldehyde, undergoes other reactions, probably with unstable chlorine by-products of the primary reaction. These unstable intermediates probably occur in amounts varying with R,, and in reacting with chlorite, probably account for the relatively large amounts of stable products, such as chlorine dioxide. The latter has served as a measure of chlorite consumption in photometric methods (7, 10) and in this study (expennients not shown) yielded variable R , values. If all oxidation states of chlorine above chloride (Cl-) were included in the analysis, and not just chlorite as usual, then the only decrease in oxidizing titer would be that equal to the organic oxidation, which if limited to the conversion of aldehyde to carboxyl, would be 2 equivalents per mole of aldehyde. Accordingly, the total oxidizing titer was determined in systems like those in Figure 1, by a procedure developed for this purpose (12). The results for benzaldehyde, arabinose, and glucose shom-ed that, respectively, 2.48, 2.42, and 2.19 equivalents of oxidizing titer instead of 2.00 were consumed per mole of aldehyde. Thus, the stated objective of this section !vas not R ttained. Consumption of total oxidizing titer greater than 2.00 per mole of aldehyde can be clue, presumably, only to failure of the method for total oxidizing poi\-er t o include perchlorate, the remaining likely form of chlorine. or to the oxidation of nonaldehydic structures. A definite choice between these two explanations cannot be made a t present, hecause no method for measuring per( hlorate under these conditions is :tvailable, b u t the writers believe that circumstantial evidence favors the fornier. The occurrence of chlorine dioxide ;tc an oxidation by-product makes the expectance of other oxidation states uf chlorine higher than chlorite seem reasonable. The occurrence of chlorate VOL. 31, NO. 8, AUGUST 1959
1389
and chlorine dioxide as decomposition products of chlorous acid is well known (1, 3, 11). This may, however, have no bearing upon the oxidation reaction, because perchlorate from decomposition is practically ruled out. Typical decomposition titers, measured with a precision of 0.170, showed no discrepancies (12). K i t h reference to the latter explanation, the leveling off of curves in Figures 3 and 4 shows that oxidation ceases and that the remaining (20 to 280% excess) chlorous acid does not react with the relatively numerous carboxyls, hydroxyls, or other groups still present after the aldehyde has been oxidized in benzaldehyde, arabinose, and dextran dialdehyde. Also. the aldonic acids of rhamnose, mannose, lactose, melibiose, and glucose xere stable toward chlorous acid ( 7 ) . Thus, it would have t o be assumed that nonaldehydic structures are attacked in a process involving unknown intermediates existing only while the primary reaction is in progress. This assumption would presuppose that in such a process the aldehydic group would become more resistant to attack than the nonaldehydic, contrary t o the behavior of aldehydes toward all known oxidizing agents. Furthermore, regardless of the diverse nature of the aldehydes studied, it would be necessary to assume that the various residues of nonaldehydic oxidation would not be subject to further slow oxidation, thus preventing the leveling off of the curves. DISCUSSION A N D ANALYTICAL APPLICATIONS
It is ironical that chlorite, a reagent with such favorable characteristics for reaction with aldehyde groups, should have the important shortcoming of variable stoichiometry in spite of its apparent straightforward oxidation of aldehyde to carboxyl. Under the conditions thus far investigated it appears that chlorite has no fixed stoichiometric relationship to any aldehyde because the observed relationship includes one or more other reactions of chlorite, perhaps not related t o the one of interest. These other reactions vary in rate with the experimental conditions. However, these reactions are governed solely by conditions which can be fixed, and thus, reproducible R, values may be expected. This is the basis for a practical application of chlorite to aldehyde analysis. The three sets of experimental conditions used in the rate studies may be used as analytical procedures. The concentration, Ao, of aldehydes or aldehyde end groups may be calculated from thiosulfate titers using Equation 5:
Ao,moles per liter 1390
[
= 1
-
21
ANALYTICAL CHEMISTRY
(5)
This is a transposed form of Equation 2 with the same symbols, and is more suitable for analytical determinations than for rate curves. The most sensitive conditions, 0.511.1, 50" C., are useful d o n to Aa 3 X 10-6d1, but application is indicated only for known aldehydes, such as those chromatographically separated and qualitatively identified, and whose R curves are available. Under these conditions, the highest chlorite acruracy may he expected. I n the case of a polysaccharide n-hose R value cannot be detrrmined, an approximation thereof is that of a known aldehyde having a similar rate of oxidation. The least sensitive 0.5M, 25" C. procedure is indicated for those substances, apparently rare, which are unstable in the 50" C. system. Because this procedure also offers the greatest range in rates, it would be indicated for the determination of very fast aldehydes, such as benzaldehyde, dextran dialdehyde, starch dialdehyde, and perhaps other dialdehydes, in the presence of slow aldehydes such as glucose, cellobiose, lactose. and maltose. I n this connection, it \Till be feasible to select conditions from Figures 3 and 4 for the determination of fast aldehydes in the presence of slo~vones, without concurrent rengcnt decomposition. For an unknoiyn aldehyde or an unresolved mixture of aldehydes or polysaccliaridcs the w e of the 3M proccdurc is indicated bwause the narrow range of R , values allows the use of a mcan R . The use of the mean R,, 2.73. the arithmetic mean of values of Experiments 21 to 29, respectively, however, nould lead to unnecessary error a t the convenient overnight reaction time of 16 hours. Because of curve recrssion the mean 16-hour ordinate is 2.53 (derived from the Rle values 2.43, 2.45, 2.46, 2.46, 2.49, 2.57, 2.66, 2.67, and 2.56 from the same experiments in the same order) with a mean deviation of 2.6%. Thus, the uncertainty in determining amounts at a level A. = 6 X 10-5MJ of unknown aldehydes lying within this range of R , should be approximately 2.6% when applying the mean R factor to 16-hour results of the 3.V procedure. Obviously, the narrower the range of R, (if, for example, the 31M procedure were applied to aldehydes faster than rhamnose), the smaller would be the uncertainty in the use of a mean R. The 3-44 procedure may be used for a known aldehyde with a specific R, obtained from a curve plotted from results using known quantities of the same aldehyde. The shorter reaction times of the 3M procedure may be preferable to higher precision. The hours required to reach maxima for the 3 M , the 0.5MJ
50" C. and the O.5M, 25" C. procedures for three sugars were as follows: arabinose, 2.9, 3.5, 5.3; rhamnose, 6.5, 13.5, 25; glucose, 15,35,49,respectivrly. The precisions for the three procedures are in approximately inverse proportion to the chlorite concentrations. These R values were primarily intended for subsequent work in this laboratory with polysaccharides, but should be applicable by others. However, confirmation for all aldehydes, with the possible exception of the much-studied glucose in 0.5M phosphate, is desirable, because the present aim as more extensive than intensive, and only in this n-ay vi11 general applicability be established. Although variable ratios affect accuracy, it is felt that the use of a weakly acidic medium avoids the much more fundamental types of uncertainty inherent in methods which use alkaline media to achieve reactivity of aldehyde groups. ACKNOWLEDGMENT
The authors thank Allene Jeanes for supplying dextran dialdehyde and information about it and Frank E. Young for the sample of D-xylose. LITERATURE CITED
(1) Barnett, B., Ph.D. dissertation, Uni-
versity of California, Berkeley, Calif., 1935.
( 2 ) Davidson, G. F., Nevell, T. P., J . Tertilelnst. 46, T 1407 (1955). (3) Jeanes, Allene, Isbell, H. S., J . Research 9atl. Bur. Standards 27, 125
(1941). (4) Jeanes, Allene, Wilham, C. A,, J . Am. Chem. Soc. 72, 2655 (1950). (5) Launer, H. F., Tomimatsu, Toshio, ANAL.CHEM.26,382 (1954). (6) Launer, H. F., Tomimatsu, Yoshio, J . A m . Chent. Soc. 76,2592 (1954). (7) Launer, H. F., Wilson, W. K., Flynn, J. H., J . Research h'atl. Bur. Standards 51 , 237 (1953). (8) Launer, H. F., Wilson, W. K., un-
published results.
W.,Alexander, B. H., Lohmar, R. L., Wolff, I. A., Rist, C. E.,
(9) Sloan, J.
J.Am. Chem. SOC.76,4429 (1954). (10) Stitt, Fred, Friedlander, Stanley, Lewis, H. J., Young, F. E., ANAL. CHEM.26, 1478 (1954). (11) Taylor, hl. C., White, J. F., Vincent, G. P., Cunningham, G. L., Ind. Eng. Chem. 32, 899 (1940). (12) Tomimatsu, Yoshio, Launer, H. F., ANAL.CHEX. 29, 1500 (1957). (13) Tomimatsu, Yoshio, Launer, H. F., unpublished results.
(14) Wilson, W. K., Padgett, A. A., T a p p i 38, 292 (1955).
RECEIVED for review January 28, 1958. Accepted February 24, 1959. Division of Carbohydrate Chemistry, 133rd Meeting, ACS, San Francisco, Calif., April 1958. Mention of specific products does not constitute endorsement by the U. S. Department of Agriculture.
