328 RIDYARD: A STUDY OF THE DETERMINATION [Analyst, Vol. 91 A Study of the Determination of Thiamine in Breakfast Cereals BY H. N. RIDYARD* ( T h e Reseavch .4 ssociatioii of British Flour-Millers, Cereals Research Siatiov, Old 1.ondon Road, S t . A lbans, flevts.) The method of determining thiamine that involves the purification by base-exchange on sand, is satisfactory for materials of the “breakfast cereal” type. Traces of materials responsible for errors in the direct determination remain after treatment, but the errors are greatly reduced and oppose one another, so that they may reasonably be neglected. PROCESSES such as pressure cooking and subsequent baking, when applied to many cereal foods, result in the destruction of thiamine. At the same time, materials are produced that give a high fluorescence on treatment with sodium hydroxide and subsequent extraction with isobutanol (“blank fluorescence”l), and also substances that interfere with the fluorescence of thiochrome.2 These effects rendered the “direct”l determination of thiochrome in unpurified extracts of such foods of little value until a more detailed study was made.The “blank” used was usually slightly greater than the total fluorescence that was developed after oxidation of the thiamine to the thiochrome, and although the blank and interferences opposed one another the relative magnitudes of the effects were unknown. Recently, the addition of synthetic thiamine to such products during manufacture has become common practice, and a more detailed and prolonged study of one such product has been made which has given interesting results.Confirmatory results of a much less detailed nature have been obtained with five other products. METHOD APPARATIJ S- were used. The apparatus used has been previously de~cribed.~ y 4 In addition, 100-ml conical flasks REAGEXTS- Extract-This is prepared as described for flour,l but with 30 g of the product in a 1000-ml conical flask and 750 ml of 0.2 N acid. Large amounts are used to compensate for the probable uneven distribution of thiamine; 30 g is a convenient approximation to 1 oz, and the rate of fortification of flour is commonly expressed in terms of mg per oz. The extract is filtered next day through a large fluted filter-paper. Thiamine additions for recovery ex$erimenfs-Place 5 ml of a 200 pg per ml solution of thiamine (stage 1 in preparation of standards1) in a 100-ml calibrated flask and fill to the mark with the extract.Mix the solutions by emptying the flask turbulently, inverting it in a dry 200 or 250-ml conical flask equipped with a glass stopper. Shake the flask, as vigorously as is possible without the formation of a lasting foam, for about 3 minutes. Return the liquid to the measuring flask, and repeat this process at least twice more. Dilute 10 ml of this solution to 100 ml with the extract. Mix the solutions as before, then dilute 10 and 20 ml of this solution to 100 ml with the same precautions, to give extracts with +0.1 and +0.2 pg per ml of added thiamine. REAGENTS, STANDARDS AND PROCEDURE- The same as those used for the “direct” method as described.l The same as those used for the “sand” method, as previously de~cribed,~ omitting those that were only used for digestion (unless this is to be undertaken).* Present address: “Silverwood,” 55 Bucknalls Drive, Rrickctt IYood, Watford, HertsMay, 19661 O F THIAMINE I N BREA4KFAST CEREALS 329 RE s u LTS At first, recoveries with the sand technique seemed disappointingly erratic but, after four sets had been examined (eight additions), it was found that the mean recovery in the eight results was 98.8 per cent., although extreme results were 108 per cent. and 90 per cent. One difficulty in recovery experiments is that each result expresses the sum of two errors of deter- mination, and it was later decided that the mixing of solutions was still a source of error, in spite of the stringent precautions laid down previously, and hence the shaking in a conical flask, as described above, was prescribed as an additional precaution.Duplicate eluates, washed with 3 portions of isobutanol, gave a higher mean recovery, but the results were not less erratic. Examination of these earlier results suggested that both fluorescent and interfering substances were held on the column, probably with varying tenacities. Some were possibly held by base exchange, others only by loose adsorption. To test this hypothesis a large uniform bulk of extract was prepared and, to portions of this, additions of thiamine were made in the manner described above. A set of 8 sand columns was prepared for each solution (+0.0, +0*1, $0.2 pg per ml of added thiamine) and 10 ml of solution were applied to each column (a total of 24 columns). Four columns of each set were then washed with 200 ml of 0.2 x hydrochloric acid, and the remainder with 500 ml of this acid.All the columns were then eluted, giving six sets of eluates. Two eluates from each four were then washed with 4 portions of 25 ml of isobutanol. The eluates were then arranged in the order 1,2,3,4, corresponding to Table I, oxidised and extracted with four portions of isobutanol and the fluorescence measured on 3 successive days as already de~cribed.~ With the isobutanol-washed eluates, the fluorescence found was multiplied by a factor of 1-09 to allow for the increase in volume of the isobutanol owing to the smaller loss of isobutanol into the isobutanol-saturated aqueous phase.The fluorescence of the washings was measured separately. The results are given in Table I. TABLE 1 EFFECT OF \VASHIKG THE COLUMN \VITH ACID AND THE ELUATE WITH ISOBUTANOL I. Eluates not washed with isobutanol 200 ml of acid-wash of column. 500 ml of acid-wash of column Order of measuring r---- A -- ~ ---L------ 7 fluorescence . . 1 3 A 4 Added B, . . . . +o.o f0.1 t 0 . 2 fO.0 +om1 +0.2 +O.O f0.1 t 0 . 2 +0*0 +0*1 +0*2 Percentage recovery - 94 89 -- 03 05 - 0 R,, pg per nil . . 0.362 0.456 0.340 0.365 0.458 0.554 0.362 0-457 0.558 0.359 0.462 0.555 95 88 - 103 98 Mean 92.75 Mean 06.0 11. Eluates washed with isobutanol B,, pg per ml . . 0.340 0453 0.547 0.355 0.454 0.548 0.355 0.4.54 0.554 0.360 0.453 0.565 Percentage recovery - 113 104 - no 97 - 99 100 - 03 103 Mean 103.25 Mean 98.75 It will be seen that the washing of the columns with 500 ml o f acid results in a slight improvement in the recovery.This indicates the slow removal of adsorbed interfering substances. The washings from the eluates, when examined in the fluorimeter, showed a remarkably constant level of 0.026 pg per ml with a maximum deviation of 0.001,, mean 0400,. The blank was reduced to 0.017 on eluates from both the 200-ml and 500-ml acid-washed columns. The blank on the isobutanol-washed eluates was reduced to 0-007 pg per ml. These figures support the view that the column holds traces of both fluorescent and interfering bodies with varying degrees of tenacity, and that these small variations may contribute to the erratic recovery.iVashing the columns with 500 ml of acid will improve the recovery but it will give much the same basic level, because of the balancing opposition of blank and interference. In order to demonstrate still further the validity of these concepts, the isobutanol extracts were concentrated by distillation i n V ~ C U O by using an oil pump. Solid carbon dioxide and alcohol were used as a condensing refrigerant, and a trickle of carbon dioxide was passed from a Kipp's apparatus into the distilling liquid to prevent bumping and also to render330 KIDYARD: A STUDY OF THE DETERMINATION [AIZazyst, Vol. 91 the alkali less soluble in the isobutanol. When the contents of the distillation flask had been reduced to a paste, it was stirred with dry isobutanol and spun in a centrifuge.The super- natant liquid was poured into a small flask. The solid was extracted three times in all in this way, and the combined extracts were then distilled again amost to dryness. The distillation residue was extracted with 4 portions of 0.5 ml of dry isobutanol, which were then transferred to an ignition tube and stored in a desiccator containing a small beaker of dry isobutanol to maintain a saturated atmosphere. The liquid was then chromatographed on paper pre- viously extracted with flowing wet butanol for 48 hours and dried. The chromatogram was developed with water-saturated hutanol in descending flow. The isobutanol extract of a 200 pg per ml thiamine solution was used as a marker without concentration. The chromato- grams when developed and dried were exposed to ultraviolet light and the fluorescence photographed.Fig. 1 is composed from two such chromatograms, and shows the results of chromatographing extracts from: ( I ) , autoclaved wheat ; (11), autoclaved wheat after the addition of thiamine ; (111), autoclaved wheat after final toasting ; (IV), pure thiamine. In order to show more clearly the nature of the interferences remaining after sand purification, 500 ml of the isobutanol washings of the eluates before oxidation were concen- trated to 2ml without being spun in a centrifuge as described above. The chromatogram of the liquid is shown in F-ig. 2, A l ; that of the liquid after stirring in some solid is shown in Fig. 2, A2. Fig. 2, A3 is the chromatogram of a pure thiochrome marker, and A4 is the washing of an eluate from sand alone.The chromatogram shows the presence of fluorescent material (blank) in the washings of the cereal eluates. The subsidiary spots on the pure thiochrome line are attributed to a repeating pair of faint spots that were noticed in such runs from time to time, and have not yet been explained. The chromatogram was then sprayed with the highly fluorescent isobutanol extract of an oxidised thiamine standard containing 200 pg per ml, and then photographed with a shorter exposure. Thus the quenching effects are developed more clearly, as in Fig. 2, B. The same chromato- gram was then sprayed with dipicrylamine to show the location of potassium on the paper (Fig. 2, C.) and finally with potassium ferrocyanide to show the location of ferrous iron (Fig.2, D). These ions are important, not for themselves but as carriers of the chloride ion which is a notorious quenching agent. The four photographs together show the quenching on the paper due to each ion, and also the presence of discrete spots which are presumably due to the presence of organic components from the washings. These are the quenching components which persist throughout the determination. The presence of both blank fluores- cence and quenching materials are thus shown clearly. I t can be seen that traces of iron in the washing from the eluates of the breakfast cereals run further than in the sand blank, where they remain at the origin. This appears to be due to the presence of organically combined iron. In the isobutanol extracts from oxidised eluates, iron is completely, or almost completely, removed by precipitation with sodium hydroxide, and the potassium (or sodium) chloride concentration is reduced by greater aqueous dilution. However, traces of potassium or sodium chloride will remain, but the quenching that arises from this is extremely small, and acts as compensation to the blank.I t has been found with this particular product, after frequent examination for about two years by both the direct and sand methods, that the direct method gives much the same result as the sand method if th.e value for the blaizk is izot deducted. This was observed previously with uncooked wheat p r ~ d u c t s , ~ ? ~ and it is interesting that it should be true with the much higher blank obtained with this cooked material.In the examination of 63 extracts of 52 samples of one wheat product it was found that the blank had a mean value of 0.083 pg per The dark spots near the origin are due to the quenching of the fluorescence. Fig. 1 . Fig. 3. Fig. 2 . Chromatograms of extracts from: ( I ) , autoclaved wheat; (11), autoclaved wheat after the addition of thiamine; (III), autoclaved wheat after final toasting; (IT), pure thiamine Chromatograms o f isobutanol extracts : E, pure thiamine; F, G, H, foods prepared from wheat; J, food prepared from rice; K, food prepared from maize Chromatograms of jsobutanol washings of eluates before oxidation: A, run in water-saturated jsobutanol ; B, sprayed with thiochrome ; C, sprayed with hexanitro-diphenyl- amine (dipicrylamine) ; D, sprayed with potassium ferrocyanide after spraying with dipicryl- amine.I . Eluate washings, solution alone; 11. solution and solids; 111. 200 pg per ml of B; IV. washing of eluate from sand aloneE F G H J K Fig. 3May, 19661 OF THIAMMIIVE I N BREAKF-AST CEREALS 331 ml (the maximum value was 0-120 and the minimum value 0.060). Where the thiamine content was in the region of 0.4 pg per ml, the value, as determined by the sand method, was on the average 0.015 pg per ml higher than that determined by the direct method (mean of 2) witliozd deductioit q f f h e blank value. The maximum differences were +Om046 and -0.022 pg per ml. In three tests, however, when the thiamine content rose to 0.5, 0.6 and 0.7 approximately, the corresponding excesses of the sand values were 0-030, 0.087 and 0.133.I t appears likely that with high interferences the quenching is proportional to the square of the thiamine concentration, as with Therefore, at certain concentrations of this material, a measure of the dircct value without deducting the blank value could be a valuable rapid routine check on the thiamine content. However, the balance of errors is not the same with other foods, even if they are prepared from wheat (see Table 11). The matter should be carefully checked by the sand method with manjT samples. TABLE I1 THIAMIKE DETERMINATION OK BREAKFAST CEREALS 30 g per 750 ml acid extracts Cereal : B8619 RIaize B8620 IVheat B8623 Wheat I38624 Wheat B8625 Rice Direct determination- pg per ml .. . . . . 0-289 0.326" 0.11 0*101* 0.093 0.075 0.311 0.290* Blank . . . . . . . . 0.105 0.092 0.070 0-055 0.070 0.070 0.070 0.08," - 0.396 Addition + 0.1 pg per ml . . 0.408 0.20 7 - - 0.460 Alddition + 0 2 pg per ml . . 0.492 0.277 - mg per 100 g blank deducted 0.46 0.56 0.11 0.12 0.06 0.01 0.60 0.52 mg per 100 g blank not deducted 0.72 0.79 0-28 0.26 0.22 0.19 0.78 0.72 Sand determination- pgpqc: ml . . . . . . 0.230 0.2l9 0.057 0.073 0.072 0-030 0.284 0.209 Addition + 0.1 pg per ml . . 0.345 0.165 - Addition + 0.2 pg per ml . . 0.425 0.263 - II'ashed eluates, pg per in1 . . 0.212 0.0s 1 - - - IVashings, pg per ml . . . . 0.041 0.035 - - 0.30; - 0--110 - - mg per 100 g . . . . . . 0-57 0.62 0.28 0.18 0.18 0.08 0.71 0.51 Manufacturers claim, mg per 100 g 0.60 0.60 * Figures in italics were obtained from difierent extracts of the sample, and lack of corre- spondence of these values with the first determinations are indicative of an erratic distribution of vitamin in the original sample.The value of taking a large amount (30g) for the extract was shown by examining duplicate extracts of 26 samples of one brand of cereals for 15 months. Mean deviations of the results from means of their respective pairs was 0.02 on a general level of 0-9 mg per 100 g, with a maximum deviation of 0.07. Six extracts of each of 5 samples were prepared with only 2 g. Mean deviation from the means was 0.07 mg per 100 g with a maximum of 0.17. Two 5-g portions of a single sample of one fortified breakfast cereal were digested, purified by base exchange on Decalso, and the thiamine determined bv using the method des- cribed by the Aneurine Panel of the Analytical Methods Committee.6 The mean of the t1z.o results obtained was 0-86 mg per 100 g.The same fluorirneter readings interpreted from a curve prepared from five standards, instead o f by calculation from a single 0-2 pg per ml standard as prescribed by the Analytical Methods Committee method, gave a mean result of 0.93. The mean value obtained from two 30-g portions examined by the sand technique was 0.95 mg per 100 g. The low result by the above calculation is due to variation with concentration of the response of the fluorimeter used, as is shown by the sigmoid form of the calibration curve.] Thus, if the calculation is performed with the reading from the 0.4 pg per ml standard in the five mentioned above, a mean value of 0.90 is obtained.Single samples of five other brands of "breakfast cereal'' have been examined by both the direct and sand methods. As one of these brands was a wheat product that was believed to have had only part of the cooking and treatment given to the brand most studied, and two of the other brands were maize and rice products, respectively, these three were examined a second time with additions, and the results from these are included in Table I1 (the columns in italics). As the distribution of thiamine when added to these prodccts t c r d s to be erratic, repeat figures on different extracts are valueless as The results are given in Table 11.332 RIDYARD [A4Tzazyst, Vol.91 a guide to reproducibility of the method. The recovery results are the best guide to this, as these are based on one extract with and without additions. The isobutanol extracts from the original determinations of these five products were concentrated and separated by chromatography. Fig. 3 has been composed from the two chromatograms so prepared, and shows: E, pure thiamine; F, G, H, foods prepared from wheat; J, food prepared from rice; K , food prepared from maize; the last two being fortified in manufacture. It is concluded that the method involving the purification by base exchange on sand is satisfactory for cereal foods that have been manufactured by pressure cooking and subsequent baking or toasting. Traces of fluorescent material are held on the column, apparently by base exchange. Traces of fluorescence-quenching substances and possibly also traces of light-absorbing materials, which may be the same as the quenching substances,2 are held on the column in a less firm and reproducible manner. These last effects oppose the blank however, and both being small are advisedly neglected, as attempts to remove them by washing the eluates with isobutanol are somewhat tedious and liable to give rise to small errors owing to partition and volume effects. I t was found that the procedure previously laid down for mixing solutions prepared in measuring flasks,l stringent though it was, was quite inadequate for the somewhat viscous solutions, and this appears to have been the main cause of erratic recoveries. Hence the more extreme methods laid down in this paper were found to be essential. The direct method may have value as a rough routine check, but the balance of blank and interference varies according to the treatment and nature of raw material. I thank Mr. K. H. IVillis for carrying out the analyses and the photography. REFERENCES 1. 2. 3. 4. 5. 6. Ridyard, H. N., i3 izalvst, 1949, 74, 18. - , Ihzd., 1030, 75, 634. - , Ibzd., 1961, 86, 723. -, Ibztl., 1!340, 74, 24. __ , J . SOC. Cliem. I n d . , 1946, 65, 92. Analytical Methods Committee, A nalyst, 1951, 76, 127. Received Novenabev 225~2, 1963