Determination. of Thiamine and Riboflavin in the Presence of Reduced Iron ELMER DE RITTER
AND
SAUL H. RUBIN, Nutrition Laboratories, Hoffmnn-LaRoche, Inc., Nutley, N. J .
Reduced iron (ferrum reductum) causes partial destruction of thiamine and riboflavin during acid extraction of certain vitamin-mineral concentrates, such as flour premixes, in which relatively large amounts of iron are incorporated. The effect is not seen in flour which has been enriched with this type of premix nor in premixes containing ferrous or ferric iron. The effect can be circumvented either by extracting at room temperature with acetate buffer at pH 6 or, if hot acid extraction is required, adding cystine to the acid extractant.
R
LOSS O F THIAMINE AND RIBOFLAVIN DURING ACID EXTRACTION O F FLOUR P R E M I X E S
EDUCED iron (ferrum reductum) is a common source of iron for enrichment of flour, corn meal, and other food products. It is usually added to these foods in the form of a highpotency premix. Other forms of iron which are commonly used for dietary supplementation include ferrous sulfate, ferric pyrophosphate, and sodium ferric pyrophosphate. I n view of the frequent combination of iron and vitamins, both in foods and pharmaceuticals, the influence of the various forms of iron on the determination of the vitamins is of considerable importance. I n a study of the quenching effect of electrolytes on the fluorescence intensity of thiochrome and riboflavin, Ellinger and Holden ( 6 ) have shown that both ferrous and ferric iron exert quenching effects which are proportional to the concentration of iron. Calculations from their data show, however, that the concentrations of ferrous and ferric iron employed in their experiments far exceed those encountered in the fluorometric analysis of the materials discussed in the present paper. Sealock and Goodland (18) have pointed out slight interference by ferric chloride in the diazotized aminoacetophenone (red dye) reaction for thiamine a t a low ratio of iron to thiamine. By analogy with the inhibitory effect of mercuric chloride, they suggested that the extent of interference depends upon the relative concentrations of interfering substance and thiamine. On the other hand, Auerbach ( I ) has found it possible to determine thiamine in pharmaceutical preparations by the red dye reaction in solutions containing unspecified amounts of ferric ammonium citrate. Interference by reduced iron in thiamine and riboflavin assays of vitamin-iron concentrates has not, however, been reported heretofore. That such interference dpes occur is demonstrated in the present report. Various factors bearing on this effect have been studied and two methods of circumventing the effect have been developed.
I n the analysis of flour premixes containing reduced iron, low and erratic thiamine and riboflavin values were obtained when these premixes were extracted in the usual fashion with 0.1 N sulfuric acid a t 100' C. (Table I). In order to evaluate the role of the acid, comparative trials were made between hot acid extraction and extraction with water by simply shaking a t room temperature. Thiamine was determined in direct extracts (no . enzyme or Decalso treatment) by both the thiochrome method (9) and the red dye reaction ( I O ) , and riboflavin by both fluorescence (26) and microbiological assay (8, 19). Table I shows typical comparative assays of flour premixes containing reduced iron or sodium ferric pyrophosphate. Replicate assays a t different times, of premixes containing reduced iron, often gave results which were in poor agreement. As is shown below, the erratic results are associated with casual variations in the technique of preparing the extract. In these particular runs, both thiochrome and red dye values for thiamine were 20 to 26y0 lower after acid extraction than after water extraction for the premixes containing reduced iron. For the premix containing sodium ferric pyrophosphate, however, both extraction procedures gave the same results a t expected levels. The red dye values for premixes containing reduced iron were definitely lower than the thiochrome values, whereas both the thiochrome and red dye methods were in agreement for the premix containing sodium ferric pyrophosphate. It was shown by Decalso purification that the difference between thiochrome and red dye values in the former case is due to inhibition of the red dye reaction. -4 greater loss of riboflavin, 32 to 40%, occurred during hot acid extraction of premixes containing reduced iron, and again, no loss was found for the premix containing sodium ferric pyrophosphate. Good agreement was obtained in all cases between fluorometric and microbiological assays when these were Table I. Extraction of Thiamine and Riboflavin from Flour Premixes carried out on the same extracts. I t was Flouf Water.4 0.1 N H t S 0 4 b Water 0.1 S H&Or Premix T y p e of Extraction Extractlon . DifferExtraction Extraction Differdemonstrated by means of an increment NO. Iron a t 24' C. a t 100' C . ence a t 24' C. a t 100' C. ence technique that the low fluorometric reThiamine sults after acid extraction in the presence Thiochrome method Red dye method of reduced iron are not due to quenching .Ilg./Ounce .lfg./Ounce 70 .lfg./Ounce .\fg./Ounce % of riboflavin fluorescence by iron or other 1 Reduced iron 381 296 22 340 267 21 Reduced iron 660 530 20 608 448 26 2 constituents of the premix extracts. The 3 Reduced iron 744 562 24 676 508 25 Sodium ferric 4 fact that both fluorometric and micropyrophosphate 443 432 2 441 446 -1 biological assays were lowered equally by Riboflavin acid extraction, supports the viex that Fluorometric method hlicrobiological method riboflavin is actually destroyed. Prolonged 1 Reduced iron 234 150 36 232 I50 35 2 Reduced iron 424 287 32 430 284 34 aeration of these acid extracts caused no 3 Reduced iron 278 4.0 460 282 39 460 . increase in the fluorescence, indicating that 4 Sodium ferric pyrophosphate 232 228 2 236 23 1 2 this destruction of riboflavin is not simply a Shaken mechanically f o r 45 minutes. b Heated 45 minutes in boiling water with occasional shaking. a reduction to leucoflavin, since the latter is easily reoxidized by air to riboflavin. 243
ANALYTICAL CHEMISTRY
244
obtained in the presence of both the .ferrous and ferric compounds, but slight losses were found in the presence of reFlour duced iron, as shown in Table 111. These Premix Water 0.1 N HzSOi HnO 0.1 N HaSOc losses are considerably smaller than those NO. Extraction Extraction Difference Extraction Extraction Difference Thiamine found under comparable conditions (cf Thiochrome method Red dye method Table 11) in the extraction of flour preMg./Ounce Mg./Ouncs 76 ,kQ./Ounca Mg./Ounce % mixes. Possible reasons for this difference 6 Reduced iron 720 3486 52 639 308 52 are discussed below. 6 Reduced iron 354 1876 47 322 156 52 7 Sodium iron Effect of pH. In order to arrive a t B pyrophosphate 348 358 -3 350 352 -1 better understanding of this effect of reRiboflavin duced iron on thiamine and riboflavin durFluorun.etric method Microbiological method ing acid extraction, a study was made of 6 Reducediron 469 158 66 444 84 81 6 Reducediron 218 68 69 215 38 82 the influence of pH, time of shaking, and 7 Sodiumiron iron-vitamin ratio. In studying the effeci pyrophosphate 231 225 3 234 227 3 of pH, McIlvaine's phosphate-citric acid By mechanical shaking for 45 minutes. b Aasay b y fermentation 411 mg. per ounce. buffer series (6) was selected in order 6 Assay b y fermentation 202 mg. per ounce. to cover a wide range, from pH 2 tc. 8, with the same buffer ions. Acetate buffer was also studied over the range of pH 3 to 6.5, in order to determine whether the specific buff& When these premixes were extracted by shaking in acid a t influenced the reaction. room temperature, the loss of thiamine, as judged by comparison with aqueous extraction, was even greater in the presence of Two milligrams of thiamine in 20 ml. of 0.1 A' hydrochloric reduced iron than at 100' C., both thiochrome and red dye acid, 1.5 mg. of riboflavin in 20 ml. of 0.05 N sulfuric acid, and methods giving losses of about 50%. Assays of these extracts 20 ml. of phosphate-citric acid buffer a t the desired p H wert by the fermentation method of Schultz, Atkin, and Frey (17) mixed and titrated to that pH with 0.1 N sodium hydroxide in also gave low values (Table 11), providing further evidence by the Beckman pH meter. Two hundred milliliters of water and 50 mg. of ferrum reductum were added and the flasks were an independent method of the destruction of thiamine. Riboshaken for 30 minutes a t room temperature. The same proceflavin loss a t room temperature was also greater than after hot dure was followed for the acetate buffer, except that the pH was extraction. The microbiological results on the acid extracts adjusted by titrating with 2.5 $1 sodium acetate, followed bq were somewhat lower than the fluorometric, which is apparently 0.1 N sodium hydroxide if the volume of acetate solution required was more than 20 m1.-Le., a t higher pH values. After shaking. due to the presence of derivatives of riboflavin which, though thiamine was determined by thiochrome and riboflavin by direct fluorescent, are less active than riboflavin for L. casei. The fluorescence measurement. authors have previously observed such discrepancies in the relative fluorometric and microbiological response of succinyl Figure 1 shows the vitamin recoveries obtained over the pH esters of riboflavin (8). Again, there were no losses from the ranges studied. With the phosphate-citric acid buffer, low premix containing sodium ferric pyrophosphate. recoveries of both vitamins were found up t o pH 6.5, but abovr Although the data in Table 11 were obtained with premixes other than those of Table I, a similar comparison of acid extraction a t room temperature and 100" C. of the premixes of Table I has confirmed the greater losses a t room temperature. To Table 111. Effect of Iron Compounds on Thiamine and Riboflavin during Acid Extraction" determine whether the greater losses were due to the continuous Per cent Recovery mechanical shaking a t room temperature, a comparison was M g . of Fe Thiamine RiboflaviI: Mg. of Fe made of mechanical shaking of 0.150 gram of a flour premix in per Mg. of per Mg.of (thio(fluoroIron Compound Thiamine Riboflavin chrome) metric' 300 ml. of 0.1 N sulfuric acid added a t room temperature and a t 7 9.3 92 97 Reduced iron 100" C. After 20 minutes' shaking, the temperature of the 7 9.3 100 100 Ferrous sulfate 7 9.3 100 101 Ferric pyrophosphate latter dropped to 65" C. Thiamine and riboflavin losses a t room temperature were 57 and 63%, and in the hot acid 28 0 3.0 mg. of thiamine and 2.25 mg. of riboflavin in 150 ml. of 0.1 N HISO. ahaken 15 minutes a t room temperature with indicated iron compound. and 50oJ,, respectively. It is evident that even with the same shaking, greater losses occur a t room temperature, which suggests that these differences bear an inverse relation to the rate of solution of the iron and vitamins. A possible mechanism is 100 discussed below. 90 90 Effect of Decalso Treatment. Aliquots of the 100' C. acid I ,',/' extracts were passed through Decalso columns, using the Decalso technique of Hennessey (9) for the thiochrome and of Hochberg et al. (IO)for the red dye method. In neither case did the Decalm purification increase the assays above the thiochrome values on the untreated extracts. Hence, it appears that the low values are not due to inhibition of fluorescence or color reaction, but to partial destfuction of thiamine during acid extraction. 30 Influence of Ferrous and Femc Iron. The effect of two commonly used iron compounds, ferrous sulfate and ferric pyro20 - -- ACETATE BUFFER 0 a 0 THIAMINE phosphate, on thiamine and riboflavin during acid extraction a t 10 PHOSPHATE-CITRIC ACID BUFFER A a A RIBOFLAVIN room temperature was studied in model experiments by shaking acid solutiona of the vitamins with these compounds. For comparison, a similar test was made with reduced iron, the ironvitamin ratio being approximately the same in each case as in flour premixes. Complete recoveries of both vitamins were Table 11. Extraction of Thiamine and Riboflavin from Flour Premixes. at 24' C.
T?t$f
-
--
.
I
-
'
V O L U M E 19, NO. 4, A P R I L 1 9 4 7
245
70 -
* 60-
B
---
I N O.IN H,S04
40-
- IN pH 5.0 PHOSPHATE-
30-
CITRIC ACID BUFFER 0 OTHUMINE A ARlBOFLAVlN
’
20-
I
101
I
5
0
10 I5 SHAKING TIME
20
- MINUTES
25
Figure 2. Effect of Time of Shaking on Destruction of Thiamine and Riboflavin by Reduced Iron
6.5 a sharp rise occurred, so that from pH 6.6 to 7 recoveries were virtually complete. Above pH 7, the thiamine recovery dropped off again, probably because of the instability of thiamine in the alkaline region. With acetate buffer, the recoveries a t lower pH values were considerably higher, nearly quantitative recoveries being obtained above p H 5.5. The marked and unusual effect with the phosphate-citrate buffer can be ascribed to the reduced iron, since quantitative recovery of the two vitamins ‘we8 obtained after 30 minutes of shaking a t pH 5 without reduced iron. Because of the wide differences in vitamin recoveries between the acetate and the phosphate-citrate buffers, particularly a t pH 5 to 6.5, the solubility of reduced iron in these buffers (Table IV) was determined under the above experimental conditions. After 30 minutes’ shaking at p H 5.0, only 0.13y0of the iron had dissolved in acetate buffer, whereas 75% had dissolved in phosphate-citrate buffer. At pH 6.0, the percentages were 0.10 for acetate and 8.0 for phosphate-citrate buffer. It appears that the greater vitamin losses are associated with the greater extent of solution of reduced iron. This is further demonstrated by the fact that (1) large vitamin losses were incurred in acetate buffer a t p H 3 and 4 where the extent of solution of the reduced iron was 61 and 49y0,respectively, (2) practically no vitamin losses were found in phosphate-citrate buffer a t pH 6.6 where the iron solubility was only 0.05’%, and (3) the sharp rise in recoveries in phosphate-citrate buffer between pH 6.4 and 6.6 was accompanied by a decrease in iron solubility from 4.2 to 0.05%. That these losses are not entirely a function of the extent of solution of the iron, however, is indicated by the circumstance that the losses after 30 minutes’ shaking in 0.1 N sulfuric acid are considerably less than in p H 5 or 6 phosphate-citrate buffer (cf. Figures 1 and 2), despite the fact that the iron is completely dissolved in 0.1 11-sulfuric acid. Effect of Time of Shaking. In view of the losses incurred in shaking for 30 minutes in 0.1 N acid and phosphate-citrate a t pH 5, it seemed of interest to study the rate of loss of the two vitamins in these extractants. Under the same conditions as
Table IV. PH
Solubility of Reduced Iron in Acetate and Phosphate-Citrate Buffer Reduced Iron Dissolved Acetate Phosphate citrate
% 3.0 4.0 5.0 6.0 6.4 6.6
7.0
61 49 0.13 0.10
.. ..
%
...
75 8.0 4.2 0.05 0.02
the previous experiment, the recoveries after shaking periods of 2 to 30 minutes were m shown in Figure 2. In 0.1 N sulfuric acid, thiamine and riboflavin losses were practically equal over the entire period, the major portion of the loss occurring in the first 2 minutes. In the buffer a t p H 5, the thiamine losses were greater. than the riboflavin losses and both increased progressively with prolonged shaking. It is apparent from these curves that extraction at low p H for even as short a period as 2 minutes is not safe in the presence of reduced iron. Although the rate of solution of iron in 0.1 N sulfuric acid (40, 70, and 100% dissolved after 2, 5, and 30 minutes, respectively) is much more rapid than in p H 5.0 phosphate-citrate buffer (0.2, 1.4, and 75% dissolved in the same periods), the vitamin losses are smaller in the acid than in the buffer except in the first few minutes. Effect of the Iron-Vitamin Ratio. Under the same experimental conditions as in the two previous experiments, the effect of the iron-vitamin ratio on the recovery of thiamine and riboflavin was determined in 0.1 N sulfuric acid. Two mg. of thiamine and 1.5 mg. of riboflavin were shaken for 30 minutes at room temperature with the reduced iron. The losses of both vitamins (Figure 3) were nearly the same and increased progressively as the iron-vitamin ratio wm increased, although the over-all effect was decidedly less than in the premixes.
HTHIAMINE H RIBOFLAVIN
IO
2o
t
I 0
20
Figure 3.
40 60 ma. REDUCED IRON
BO
I 100
Effect of the Iron-Vitamin Ratio
on Destruction of Thiamine and Riboflavin
by Reduced Iron PROTECTIVE ACTION OF CYSTINE
In Simple Solutions. The foregoing evidence indicates that the low values for thiamine and riboflavin obtained after acid extraction in the presence of reduced iron are not due to interference in the terminal parts of the assay methods. It seemed reasonable to suppose, therefore, that the effect is due to partial reduction of the two vitamins. It was thought possible to prevent this reduction by adding, before extraction, some easily reducible compound which might be reduced selectively in preference to thiamine and riboflavin. Cystine has been found satisfactory for this purpose. Table V shows the effect of heating an aqueous acid solution of thiamine and riboflavin a t 100’ C. with and without cystine. Since reduced iron does not exert so great an effect in simple solutions of thiamine and riboflavin (Table 111) as in extracts of premixes, it was deemed desirable to increase the relative amount of reduced iron in these experiments in order to be able to demonstrate significant effects. The amount of cystine added was increased progressively to determine the minimum cystine-iron ratio required to provide complete protection for thiamine, which manifests f h e greater loss under these conditions. The vitamin losses in the absence of cystine, as shown by the first values in Table V, can be avoided by the addition of an excess of cystine before hot extraction. Cystine does not, however, provide complete protection for the
ANALYTICAL CHEMISTRY
246
vitamins in extractions made by shaking in acid a t room temperature, wherein, as shown in Table 11, the vitamin losses are relatively greater than a t 100" C. Hence, the use of cystine is best confined to hot acid extraction. Stoichiometrically one mole of cystine would be required per mole of iron for the oxidation of Fe" to Fe++. Actually, good recovery (95%) of thiamine is already obtained at a cystine-iron ratio of only 0.3. This is not surprising, since the molar ratio of cystine to thiamine was about 90 to 1 in this solution. The last two values in Table V show that cystine itself doesnotinterfere with either fluorometric assay. It also does not interfere with the microbiologicalriboflavin assays. The reduction product, cysteine, however, has been shown by Sealock and Goodland (18) and by the authors' observations t o inhibit the red dye reaction, which necessitates prior removal of cysteine by Decalso treatment. The fact that cystine preventi loss of thiamine and riboflavin during extraction in the presence of acid and reduced iron provides further evidence that this effect may be ascribed to reduction of these vitamins. In Flour Enrichment Premixes. As an extension of the findings described above with regard to the protective action of cystine in simple solutions, cystine was incorporated in the acid extractant used for determination of thiamine and riboflavin in flour premixes containing reduced iron. These were compared with extracts of the same premixes which were prepared by mechanical shaking a t room temperature with pH 6 acetate buffer, a procedure which hrts been shown (Figure I) to give complete recovery from simple solutions. The excellent recoveries obtained (Table VI) show that both extraction procedures are applicable to flour premixes containing reduced iron. A further comparison of the two methods of Table VI with simple aqueous extraction is shown in Table VII. The experimental conditions were similar to those in the above recovery tests. Both thiamine and riboflavin assays were 3 to 5% lower after simple aqueous extraction than after extraction by the other two methods. Although small, these differences are consistent, The differences between acetate and acid-cystine extraction are about 2% and are not necessarily significant.
Table V.
Protective Action of Cystine in Simple Solutions
Reduced
Iron
Cystine
Mo.
Me. 0 33 65 163 325 650 650
50
so
50 50 50 60 0
Cystine Molea/mole Fa
0 0.15 0.30 0.75 1.5 3.0
..
Recoverya Thiamine Riboflavinb (thiochrome) (fluorometric)
%
%
71 77 95 97 100 100 98
..
95
.... ,.
99 101
1 mg. of thiamine and 0.75 mg. of riboflavin in 170 ml. of solutionof cystine in 0.1 N heated 30 minutes a t 100' C . with occasional shaking. a
b Riboflavin added only in cases where d a t a are given.
Table VI. Vitamin Recovery in Assay of Flour Premixes (Containing Reduced Iron) by Recommended Methods
Flour Premix So. 1 2 3
P e r Cent Recovery of Added Vitamin5 Extraction with pH 6 Acetate Buffer a t Room Extraction in 0.1 A- H2SOa Temperatureb Cystine a t 100" C.C Thiamine Riboflavin Thiamine Riboflavin (thiochrome) (fluorometric) (thiochrome) (fluorometric) 98 102 100 103 95 98 98 95 98 101 98 103 Av. 97 100 99 100
+
Amount8 of thiamine and riboflavin approximately equal t o those already oontained in samples added before extractlon. b Buffer consists of 10 ml. of 0.75M sodium acetate plus 25 ml. of 0 1 -1HrSOi diluted to 175 ml. Extraction for 30 minutes with continuous shaking 0 Samples extracted 30 minutes with occasional shaking in 150 ml. of 0 1 X &so4 containing E moles of cystine per mole of iron in premix sample.
Table VII.
Comparison of Extraction Procedures for Flour Premixes Containing Reduced Iron
Flour Premix No a
Water a t 24' C.
2 3 5
674 720 720 Av. 705
2 3 5
415 444 469 Av. 443
+
pH 6 Acetate 0.1 N &SO1 Buffer a t 24' C. Cystine a t 100' C Mtliigrams per ounce Thiamine Assay (Thiochrome) 695 690 784 767 757 744 734 749 Riboflavin Assay (Fluorometric) 428 435 458 471 488 497 458 468
0 Flour premixes 2 and 3 extracted in presence of 8 moles of cystine per mole of iron, while for premix 5 ratio was 3.
In Extraction of Enriched 80% Extraction Flour. Contrary to the findings with premixes, hot acid extraction caused no loss of thiamine or riboflavin when applied to enriched flour (80% extraction). The claimed values were obtained for thiamine by both the short and the complete thiochrome procedures, for riboflavin by the complete fluorescence method (26). In Extraction of Multivitamin Tablets. Comparative assays of multivitamin-mineral tablets were run with and without cystine in the acid. The tablets studied contained approximately 0.8 mg. of thiamine, 0.7 mg. of riboflavin, and 3.34 mg. of reduced iron per tablet, the thiamine and iron being contained in outer coatings and the riboflavin in the central core of the tablets. hlultiple assays were run, each on a 10-tablet sample, to obtain a statistical average. Extractions in the presence of cystine yielded an average thiamine content, as shown in Table VIII, about 6% higher than by direct acid extraction. Statistically this difference lies in the range of probable significance. A similar comparison of riboflavin assays after extraction of whole tablets with and without cystine showed practically no difference. This is not surprising in view of the fact that the iron in the outer coating is probably largely dissolved before the riboflavin begins to be extracted. When the tablets were blended before extracting, however, the iron caused slight reduction of riboflavin, as shown by the values in the last column of Table VIII. Again, this small difference is probably significant. DISCUSSION
The avaihble evidence indicates that the effect of metallic iron is due to destruction of the vitamins during extraction and not to interference in the assay methods. The agreement among thiochrome, red dye, and fermentation values for thiamine, and between fluorometric and microbiological values for riboflavin, argues against such interference. It would be a remarkable coincidence if these dissimilar methods were affected equally by inhibitors. The fact that Decalso purification fails to bring either thiochrome or red dye assays up to the expected values and the fact that there is no quenching of the fluorescence of an increment of riboflavin added to these extracts, also argue that the effect is not due to inhibitors. The quenching of riboflavin fluorescence by ferrous iron reported by Ellinger and Holden (6) was observed a t much higher iron levels than are usually encountered in the assay of enriched products or premixes. For example, in a 0.005 N solution of ferrous iron, these authors found that the fluorescence of riboflavin was diminished h y only 5.97,. In dilutions of flour premix extracts containing about 0.1 microgram of riboflavin per ml. (the concentration a t which fluorometric readings were taken), the normality of iron is less than l/lw of this value. Similarly, their data on the quenching of thiochrome and riboflavin fluorescence by ferric iron were
247
V O L U M E 19, NO. 4, A P R I L 1 9 4 7 obtained a t relatively high iron concentrations. The present data indicate no appreciable interference by either ferrous or ferric iron with thiochrome or riboflavin fluorescence in the assay of flour premixes. Since the reaction of reduced iron with acid has a strong reducing potential and both thiamine and riboflavin are reducible, it is possible that the low values are due to reduction of the vitamins. The corollary, that the addition of an excess of an easily reducible substance such as cystine prevents this loss of thiamine and riboflavin, supports this view. It is further conceivable that metallic iron may catalyze the effect through a mechanism which involves adsorption of the vitamins, the reduction being catalyzed by the iron or traces of iron oxides. In an analogous reaction, Lipmann ( I S ) reduced thiamine with zinc dust and A' hydrochloric acid. In a further study of this reaction, Lipmann and Perlmann ( 1 4 ) observed first a transient yellowishgreen compound which was believed to result from half-reduction of the thiazole ring. After the disappearance of the color, vigorous hydrogen sulfide formation set in, indicating that hydrogenation was followed by profound degradation. The reduction destroyed the biological activity, and it was not possible to reoxidize the resulting product to thiochrome with alkaline ferricyanide. Kuhn et al. (12) first showed that riboflavin is readily reduced to leucoflavin. They carried out this reduction with hydrogen in the presence of a catalyst, zinc in acid solution, sodium thiosulfate, hydrogen sulfide in alkaline solution, . and titanous chloride. Sodium hydrosulfite also effects complete reduction to the leuco compound, which is easily oxidized to riboflavin by air, a reaction which is commonly used (11, 16, 16) in riboflavin assay procedures. With an acid solution of stannous chloride, however, Chaves and Guimardes ( 4 ) found that riboflavin was only partially reduced and could not be completely reoxidized by aeration. With sodium thiosulfate these workers observed partial reduction of riboflavin, but no reoxidation occurred on aeration. The authors have found that, after partial destruction of riboflavin by reduced iron in acid solution, prolonged aeration does not restore the fluorescence of the solution, indicating that the effect is not simply a reduction of leucoflavin. Since, a t the prevailing concentrations, neither ferrous nor ferric iron exercises any effect on the vitamin determinations, it may be assumed that the action of reduced iron ceases when it has gone into solution. It seems reasonable, also, to assume that the effect occurs a t the surface of the particulate iron which dissolves more slowly than the vitamins. If this is so, the extent of interaction with the vitamins which occurs during acid extraction may be expected to depend on such factors as relative concentrations, effectiveness of contact, and rates of solution of the iron and vitamins. These factors, in turn, are influenced by pH, temperature, time and vigor of shaking, and the physical and chemical characteristics of the analytical sample. The complex interplay of all these factors would determine the extent of loss of thiamine and riboflavin. In so complex a system, it is therefore not surprising that wide variations in the magnitude of the iron effect were observed for different test materials and for the same materials under different conditions. For example, the greater destruction of thiamine and riboflavin a t 24" C. than a t 100" C. during acid extraction of flour premixes is probably due to slower polution and longer contact of the iron a t the lower temperature. The large differences in the effect of reduced iron on thiamine and riboflsvin in phosphate-citrate buffer, as compared to acetate buffer (Figure l ) ,are apparently due to a specific influence of the buffer ions, reduced iron being appreciably soluble in phosphate-citrate buffer a t p H 5 to 6, but practically insoluble in acetate in this pH range. Beadle et al. (3) and Farrer ( 7 ) have demonstrated striking differences in the effect of various buffer anions o n the thermal destruction of thiamine, although their findings do not parallel the effects described here.
The lack of significant effect of reduced iron during acid extraction of enriched flour and the small effect noted in other cases lead to the belief that other substances present exert a protective action either mechanically, or chemically as doea cystine. This protective effect might be expected to be more pronounced for materials such as enriched flour, which contains a relatively small absolute amount of reduced iron. Two methods have been found effective in circumventing the effect of reduced iron. Shaking a t room temperature with p H 6.0 acetate buffer is applicable to finely divided products ,such as flour premixes which can be completely extracted under such conditions. The effectiveness of this mode of extraction is due t o the fact that little iron goes into solution a t this pH. When hot extraction is required, a solution of cystine in 0.1 N sulfuric acid, containing a minimum of 1.5 moles of cystine per mole of iron in the sample (or 7 mg. of cystine per mg. of iron), has been found effective. ,4 greater excess of cystine will not interfere with thiamine or riboflavin assays even a t a cystine-vitamin ratio by weight of 650 to 1. For red dye thiamine assays, the reduction product, cysteine, must be removed by Decalso treatment before reaction with the diazo compound. For t h e products tested, these two extraction procedures have been found effective. For other products, containing reduced iron, the merite of these procedures can be determined by comparative trials similar to those described here.
Table VIII.
Assay of Multivitamin Tablets Containing Reduced Iron (All values in mg. per tablet)
Tablets Whole Tablets Extracted" Blended Thiamine (Thiochrome) Riboflavin (Fluorometric) before Acid Extractiona, Assay Acid +, Acid cystineb Acid No. Acid cystine 1 0.79 0.84 0.70 0.71 0.66 2 0.78 0.79 0.69 0.67 0.67 3 0.76 0.89 0.71 0.70 0.68 4 0.84 0.88 0.69 0.69 0.66 5 0.76 0.83 .. 6 0.80 0.84 ' .. .. 7 0.84 0.85 .. .. Av. 0.80 0.85 0.70 0.69 0.67
+
..
.. .. ..
a Extraction for 45 minutes a t 100' C. in 250 ml. of 0.1 N HzSO,. b 400 mg. of cystine per 10-tablet sample (molar ratio of cystine to iron apprnximntely 3 tn 1)
RECOMMENDED PROCEDURES
Finely Divided Samples. For products such as flour premixes, which do not require hot extraction or enzyme treatment, shake the sample mechanically for 30 minutes a t room temperature in pH 6 acetate buffer. Prepare the buffer solution by mixing 10 parts of 0.1 N sulfuric acid and 25 parts 0.75 M sodium acetate. Take 1 part of this buffer solution and 5 parts of water for the extracting solution. If the p H of the extract is not in the range 5.5 to 6.5, vary the amount of buffer as required. Samples Requiring Hot Extraction. Dissolve a sufficient quantity of cystine in 20 ml. of 5 N sulfuric acid so that after dilution to 1 liter, the aliquot of the resulting 0.1 N sulfuric acid taken for extraction contains approximately 13mg. of cystine per mg. of reduced iron in the sample to be assayed. Extract 45 minutes a t 100" C. with occasional shaking. After extraction by either of the above procedures, subdilute as required and determine thiamine by the thiochrome method and riboflavin by either fluorometric or microbiological assay. The red dye method may be used for thiamine assay with the proviso that after extraction with a n acid solution of cystine, an aliquot of the extract must be passed over Decalso according to the technique of Hochberg et al. (10) in order to remove cysteine. SUMMARY
Partial and erratic destruction of thiamine and riboflavin occurs when the usual acid extraction procedures are applied to vitamin-enriched materials, such as flour premixes, containing large amounts of reduced iron. In the case of enriched flour,
ANALYTICAL CHEMISTRY
248 the effect of reduced iron is negligible. No losses have been found in the presence of ferrous or ferric iron, as tested with ferrous
aulfete, ferric pyrophosphate, and sodium ferric pyrophosphate. The effect has been ascertained for thiamine by both the thiochrome and red dye methods and for riboflavin by fluorometric and microbiological assay. The available evidence indicates that inactivation is due to reduction of the vitamins, the extent of reduction depending upon the relative iron-vitamin content and the chemical and physical characteristics of the preparation, as well as the pH, specific buffer, vigor of shaking, temperature, and time of extraction. For finely divided materials (such as flour premixes) from which thiamine and riboflavin can be completely extracted a t room temperature, the effect of reduced iron can be circumvented by extraction with acetate buffer a t a p H of 6 or slightly above. For samples which require vigorous extraction, the addition of a t least 1.5 moles of cystine in acid solution per mole of iron permits safe extraction a t 100" C. Cystine does not interfere with the assay methods even a t concentrations considerably in exceas of those normally required, but the reduction product, cysteine, inhibits the red dye reaction and must be removed by passage over Decalso. ACKNOWLEDGMENT
The authors wish to thank F. J. G. de Leeuw of the Lucidol Corporation, Buffalo, for advice and for supplying various samples. H. K. Parker of Wallace and Tiernan, Inc., Belleville, N;J., also kindly supplied several samples. Elsie Febbraro and FI W.Jahns cooperated in carrying out some of the experiments.
LITERATURE CITED
(1) Auerbach, M.E., J . Am. Pharm. Assoc., Scientific Ed., 29,313 (1940). (2) Bauernfeind, J. C.,Sotier, A. L., and Boruff, C. S., IND.ENQ. . CHEM..A N ~ LED.. . 14.666 (1942). (3) Beadle, B. W., Greenwood, D. A., 'and Kraybdl, H. R., J . Bid. Chem., 149,339 (1943). (4) Chaves, J. M.,and GuimarBes, L. R., Anais assoc. qugm. B r a d , 2, 158 (1943). (5) Clark, W. M.,"Determination of Hydrogen Ions", 3rd ed.. p. 214, Baltimore, Williams and Wilkins Co., 1928. (6) Ellinger, P.,and Holden, M.,Biochem. J., 38,147 (1944). (7) Farrer, K. T.H., Ibid., 39,128 (1945). (8) Furter, M.F.,Haas, G. J., and Rubin, S. H., J . Biol. Chem., 160, 293 (1945). (9) Hennessey, D.J., IND.ENQ.CHEM.,ANAL.ED., 13,216 (1941). (10) Hochberg, M.,Melnick, D., and Oser, B. L., Cereal Chem., 22, 83 (1945). (11) Hodson, A. Z.,and Norris, L. C., J . Biol. Chem., 131,621(1939). (12) Kuhn, R., Gyorgy, P., and Wagner-Jauregg, T., Ber., 66,676 (1933). (13) Lipmann, F.,Nature, 140,849(1937). (14) Lipmann, F., and Perlmann, G., J. Am. Chem. Soc., 60, 2674 (1938). (15) Rubin, S.H., and De Ritter, E., J . Biol. Chem., 158,639 (1946). (16) Rubin, S. H., De Ritter, E., Schuman, R. L., and Bauernfeind, J. C., IND. ENQ.CHEM., ANAL.ED.,17,136 (1945). (17) Schultz, A. S., Atkin, L., and Frey, C. N., Ibid., 14,36 (1942). (18) Sealock, R. R.,and Goodland, R. L.. J . Biol. Chem., 154, 64 (1944). (19) Snell, E.E., and Strong, F. M., IND.ENG.CHEM.,ANAL.ED., 11, 346 (1939). PRESESTED before the Division of Agricultural and Food Chemistry a t the 110th Meeting of the AMERICAN CEEMICAL SOCIETY, Chicago, Ill. Publiaation 83, Hoffmann-LsRoche, Inc.
Viscosity and Density of Four Liquefied-Gas Aerosol Solutions Containing DDT LYLE D. GOODHUE
AND
A. C. IIAZEN
U. S . Department of Agriculture, Agricultural Research Administration, Bureau of Entomology and Plant Quarantine, Beltsuille, iMd.
The viscosity and density of four typical liquefied-gas aerosol solutions containillg DDT have been measured at both 25' and 30' C. A special enclosed pipet, which was used for the measurement of viscosity, is described. The densities of three solutions were measured with a small hydrometer. The density of the fourth solution, which was out of the range of the hydrometer, was measured with a special dilatometer. The viscosities ranged from 0.00352 to 0.00180 poise and the densities from 1.041. to 1.278 grams per ml.
T
HE rapid development of liquefied-gas aerosols led to the
trial-and-error method of designing dispensers and nozzles. Consequently, there is a need for accurate study and theoretical consideration of nozzles already in use that will give data for the construction of new types of nozzles. In such studies the viscosity and density of the fluids concerned are important because of their effect on their mechanical behavior. The Reynolds number (3). which involves both the viscosity and the density, is often applied to a particular case of flow, while the Froude number ( 3 ) ,which depends only upon the density, is applied in other cases. Generally one of these values is dominant. The viscosity of the liquid becomes less important when the Reynolds number is over 3000. The Froude number may then be applied in the study of the conditions of flow. The viscosity and density measurements recorded in this paper were made on four typical aerosol solutions, the composition of which is given in Table I. Formulas 199 and 179 were used for
military purposes and are now in general civilian use. Formula 351 has a higher pyrethrum content, and No. 350 is a formula with a reduced pressure due to the addition of pentane. VISCOSITY DETERMINATION
Viscosities of liquefied gases have been determined (I?) by the use of a Hoeppler-type viscometer, but since an instrument of this type could not be obtained, the instrument shown in Figure 1, embodying the pipet principle (I), %-asdesigned and constructed in this laboratory. Test tube A was made from heavy-walled Pyrex tubing by closing one end and sawing the other end ht right angles. It is 25 cm. (10 inches) long and has a 0.1-inch wall thickness and a 0.5inch internal diameter. The frame, B, 11.5 inches long, made from 0.75 inch brass pipe, holds tube A . Windows 0.625 inch wide and 8.5 inches long n-ere cut in the side of the pipe, and threads were cut on the inside at each end. A standard S-valve, C, was fitted into a 0.375- to 0.5-inch pipe
