Evaluation of the Ferrous Thiocyanate Colorimetric Method. CHARLES D. WAGNER, H. LAWREVCE CLEVER, AND EDWARD D. PETERS Shell Decelopment Company, Emeryville, Calif.
The colorimetric ferrous thiocyanate method, first proposed by Young, Vogt, and Nieuwland, has been tested on a variety of autoxidized materials. It was found fundamentally accurate on hydrocarbon materials containing no conjugated diolefins; on diolefins the results varied considerably with time and were therefore of only empirical value. The precision was relatively poor (approximately 5 to 10% of the absolute value) but the method w-as sensitive to very small peroxide concentrations. Presence of
F
ERROUS ion has been used extensively as a reducing agent
for the determination of organic peroxides.' One of the common methods involving ferrous ion, originally proposed by Young, Vogt, and Sieuwland (?), comprises reduction of the sample in an acidified solution of ferrous thiocyanate in methanol followed by measurement of the depth of color of the ferric thiocyanate produced. Bolland, Sundralingam, Sutton, and Tristram ( 2 ) used this method with slight variations in rubber research, and Farmer et al. ( 3 ) based theoretical conclusions on results using this method. Lips, Chapman, and LlcFarlane (6) used a similar method with acetone as solvent in place of methanol. Young, Vogt, and Kieuwland ( 7 )based their claim for accuracy of the method on determinations of pure succinyl peroxide. Bolland et a ~ .( 2 ) obtained analyses close to theoretical on succinyl peroxide, dihydroxyheptyl peroxide, and cyclohexene peroxide. However, Lea ( d ) , using a method similar to that of Lips, Chapman, and PvlcFarlane ();, discovered that when air was carefully removed from both sample and reagent the results were only 10 t o 30% of those obtained without such treatment; he concluded that atmospheric oxygen and dissolved oxygen were oxidizing ferrous ion in the presence of peroxide (although little atmospheric oxidation takes place in the absence of peroxide) and, therefore, that the results included very large positive errors. In view of the conflicting reports of accurate values on known peroxides on the one hand and oxidation by air to produce high results on the other hand, experiments were undertaken in order to resolve these differences. By using the method of Bolland, Sundralingam, Sutton, and Tristram, essentially accurate results were obtained on pure tert-butyl hydroperoxide, a,a-dimethylbenzyl hydroperoxide (cumene peroxide), and tetrahydronaphthy1 hydroperoxide (tetralin peroside). With autosidized mono-olefins good agreement was obtained betn-een results by the sodium iodide-isopropyl alcohol and colorimetric methods. However, in all colorimetric analyses wherein oxygen was carefully excluded, the observations of Lea ( 4 ) were substantiated. Furthermore, it was found that no additional color developed when air was passed through the mixture of oxygen-free sample and reagent, a fact which indicates complete destruction of peroxides largely by decomposition rather than reduction. It therefore appears that dissolved oxygen is necessary in this method, either t o spccd up the reduction reaction or t o inhibit the ferrous ion-catalyzed decomposition.. As a result of these studies, the colorimetric method is recommended as the best for frequent analyses of materials containing only small amounts of peroxides. For materials (except diolefins) containing lal'ge amounts of peroxides, the iodometric method is more precise and accurate. The colorimetric method
high polymer sometimes caused interference because of insolubility and turbidity. The effect of molecular oxygen on the results, first reported by Lea, ha8 been verified. Contrary to Lea's conclusion that the molecular oxygen oxidizes ferrous ion in the presence of peroxide, the presence of dissolved oxygen is necessary t o obtain quantitative peroxide results and the absence of oxygen gives very low results. KO definite evidence of abnormally high results was found in these investigations.
is not well suited for materials containing polymer, particularlj diolefins, because of turbidity caused by insoluble high polymers ASALYTICAL METHOD
Apparatus. A Spekker photoelectric absorptiometer was used, equipped with blue KO.6 and green No. 5 filters and absorption cells allowing the light t o pass through a layer of solution 1 em thick. Reagent. Ferrous thiocyanate solution was prepared as follows: Oissolve 1 gram of C.P. ammonium thiocyanate and 1 ml df 25% by weight of sulfuric acid in 200 ml. of C.P. deaerated methanol, and shake the resulting solution with 0.2 gram of finely pulverized ferrous ammoniLm sulfate. Place the decanted reagent (prepared fresh daily) in a brown glass-stoppered bottle Procedure. Introduce 1 ml. of sample preferably containing between 0.0001 and 0.0007 milliequivalent of reactive peroxide into a 25-ml. volumetric flask. (If necessary, use 1 ml. of a suitable memanol dilution of the sample.) Fill t o the mark xith ferrous thiocyanate reagent, mix well, and fill the colorimeter cell with the mixture. At 10 minutes from the time of mixing, read the optical density and convert the value, corrected by a blank determination, to terms of concentration by comparing it with the optical density calibration curve obtained by use of corresponding knoFvn amounts of standard ferric chloride sohitions. EXPERIMENTAL
Materials Used. Samples of tetralin peroxide, cumene peroxide, ter/-butyl hydroperoxide, and 30% hydrogen peroxide and autoxidized samples of diisobutylene, 2-pentene, cyclohexene, methylpentadiene, and diethyl ether were identical with those used in experiments reported in a previous paper ( 6 ) . Applicability and Accuracy. The listed materials were analyzed by the colorimetric method and also by the sodium iodide-isopropyl alcohol method (6). Results, given in Table I, show that
Table I. Accuracy of Colorimetric Peroxide Method Compared to Sodium Iodide-Isopropyl Alcohol hlethod Material
Sodium IodideIsopropyl Alcohol Method
Apparent Peroxide Content, % lerl-Butyl hydroperoxide 99.8 Cumene peroxide 94.2 Tetralin peroxide 9.5.4 Hydrogen peroxide, 30% 29.2 Peroxide No., ;\lilliequivalents per Liter Cyclohexene 2-Pentene Diirobiitylene Diethyl ether hlethylpentadiene
980
Colorimetric Method 105 112 95 24
DECEMBER 1947
981
Table 11. Effect of Sample Size and Reaction Time on Results for Autoxidized Methylpentadiene by Colorimetric Peroxide Rlethod Sample Dilution in \~lethanol 1:500 1 : 1000 'I
Peroxide S o . , hIillieqnivalents per Litera 2:s min. 170 160
5 min. 230 230
l? min. 300 300
2.0 min. 365 360
40 min. 430
410
Referred to original sample before dilution
'rahle 111. Effect of Sample Size on Results by Colorimetric Peroxide Rlethod Sample tert-Butyl hydroperoxide methanol C yclohexene
ID
Dilution 40 2 mg./l 20.1 mg./l 1:200 .rnn
Diisobutylene "Pentene Diethyl ether
1 : 200 7 :4000
!:?I :IU
Results quoted on dilute solution of pure peroxide: original sample before dilution. '1
.
Peroxide N o . , hlilliequivalents per Liter 0 96a 0 48" 98 t o 104 108 i o 112 16 to 17 16 t o 17 62 t o 70 46 to 57 2 35 2 20 all others refer t o
~
the t n o methods agree reasonably well in most cases. Poor agreement was obtained on ether and methylpentadiene (a conlugated diolefin). I n the latter case, the reduction reaction was slox\- and incomplete with both methods, indicating that 'both disagreeing results were of only empirical value. Benzoyl perosidr ~ l s oreacted only slowly with the ferrous thiocyanate reageni Since results close to theoretical (considering the low precision of the colorimetric method) were obtained on the pure hydroperoxides, there is reason to believe that the colorimetric method ib fundamentally accurate when applied to autoxidized materials containing only isolated double bonds. Effects of Sample Size and Reaction Time. Thc hydrop e r oxides r e a c t quickly the optical density inrreases rapidly during the Cork sltt to allow gas first 2 minutes and thcrepossage after remains constant. However, the peroxides in methylpentadiene, which dre presumably of the bridge type described by Bodentlorff ( I ) , react very slody, Reogent as shown in Table 11. Sample size does not seem to influenceresults appreciably on any of the materials c Influent gos tested, as shown in Tables I1 and 111. The absence of an effect of sample size with methyl pentadiene beems odd in vievv of the pronounced effect of this variable on results for conjugated diolefins by the sodium iodide method. Effect of Oxygen Concentration. In order to remove dissolved oxygen from the sample and reagent beFigure 1. Apparatus for fore mixing, an apparatus Changing Dissolved Gas in Yimilar to that described by Reagent and Sample before I.ea ( 4 ) was used (Figure 1) Mixing
Nitrogen, from which oxygen had been removed by passage over copper a t 700' C., was passed for 10 minutes through the I ml. of sample in the cylinder and then through the reagent in the funnel. At the end of this period the reagent was allowed to mix vith the sample, and a portion of the misture was removed (with nitrogen being bubbled through the solution) via a nitrogen-filled pipet to fill a colorimeter cell, through which nitrogen was passed bv means of a special cover (Figure 2). After a glass stopper was placed in the ground joint and the stopcock closed, the tubing conveying the nitrogen was removed and the optical density of the solution determined. The third column of Table IV substantiates the observation Lea ( 4 ) that much lower results arc obtained in the absence of oxygen. As a check on the effect of the passage of gas through the materials, air was passed through the sample and reagent before mixing in exactly the same way ab for nitrogen. The second column shows that good agreement is obtained viith the results by the prescribed method except in the case of pure tetralin peroside, which is somewhat anomalous. W$en air was added after mixing the deaerated sample and reagent, no deepening of color occurred, as shown in the fourth column This demonstrates that the peroxide (nondiolefin) is completely and rapidly destroyed in the presence of ferrous ion, by both reduction and decomposition In the presence of molecular oxygen. the reduction reaction takes place much more rapidly than the decomposition reaction; in the absence of molecular I cm. oxygen the reverse is true and only a Figure 2. Colorsmall quantity of ferric thiocyanatc is imeter Cell with formed. Use of oxygen in place of niCover fof Air trogen gave approximately normal Exclusion values, verifying the conclusion that a certain amount of oxygen is necessary t o obtain quantitative results and that the effect is not one of oxidation of ferrous ion b.v molecular oxygen.
ot
.
Table TV.
Effect of Oxygen Concentration on Results bv Colorimetric Peroxide 3Iethod
Apparent tert-Butyl hydroperoxide in methanol Tetralin peroxide in methann1 29.4% hydrogen peroxide in methanol
Peroxide Content, %
94
96
63
98
68
4
23 5
...
15.4
56
6
91
15 4
Peroxide N o . , Milliequivalents per Liter Cyclohexene 120 120 20 44 48 53 11 11 2-Pentene Diisobutylene 19 21 4 4 Methylpentadiene 266 246 193 180 0.8 0.8 Diethyl ether 2.3 ...
108
In the absence of oxygen lower values are also obtained OII methylpentadiene, although the degree of lowering is less marked. The presence or absence of oxygen seems to have no influence OD the rate of reduction of those peroxides remaining after mixing, since the optical density after mixing increases slowly and at ahout the same rate whether oxygen is present or not. This sugKests that the hydroperoxides are destroyed rapidly upon mixing
982
V O L U M E 19, NO. 1 2
the reagent and sample, leaving only the bridge-type peroxides intact. ACWOWLEDGMENT
The authors n-ish to express their appreciation to Louis Lykken and John Rae for advice and suggestions, and to William Everson for assistance in the preparation of this paper. LITERATURE CITED (1)
Bodendorff, K., Arch. Pharm., 271, 1 (1933).
Bolland, J. L., Sundralingam, A., Sutton, D. A,, and Tiistram, G. R., Trans. Inst. Rubberlnd., 17,29 (1941). (3) Farmer, E. H., Bloomfield, G. F., Sundralingam, A , , arid Sutton, D. -4., Trans. Faraday SOC.,38, 348 (1942). (4) Lea, C. H., J . SOC.Chem. Ind., 64, 106 (1945). ( 5 ) Lips, A , Chapman, R. d.,and McFarlane, IT. D., Oil and Soap,
(2)
20,240 (1943). (6) Wagner, C. D., Smith, R. H., and Peters, E. D.. ANAL.CHEM.,19, 976 (1947). (7)
Young, C. A , , Vogt, R. R., and Nieun-land, J. -1..IND.EKG. CHEM.,-4s.1~. ED.,8, 198 ( 1 9 3 6 ) .
R E C E I V Ehpril D 1. 1947.
Evaluation of the Ferrous-Titanous Method CHARLES D. WAGKER, RICHARD H. SMITH,
AND
EDWARD D. PETERS
Shell Development Company, Emeryville, Calif.
The ferrous-titanous (Yule and Wilson) method has been tested on a variety of autoxidized hydrocarbons and peroxides, including some pure peroxides of a type known to be present in autoxidized materials. The method generally gives results that represent only fractions of the true values. Precise results were obtained with peroxides in mono-olefins under ordinary conditions of application of the method, but it was necessary to control the experimental conditions carefully for precise results with diolefin peroxides. Being precise, the method has empirical usefulness if the results can be correlated with properties of the material in question. The method is convenient for multiple analysis, with the single disadvantage that initial preparation and storage of the reagents are more difficult than with other methods.
I
N 1931, Yule and Wilson (9) reported that peroxides in gaso-
lines could be quickly and conveniently determined by reducing them with ferrous thiocyanate in aqueous acetone and titrating the resulting ferric thiocyanate with standard titanous solution. They used results obtained in this way chiefly as a measure of the gum-forming tendencies of the gasoline, and stated that other methods were less suitable, although they recognized that the results by the ferrous-titanous method did not represent actual peroxide content. The latter fact was evident in view of dependence of results on sample size and of the observation that certain materials treated with ferrous sulfate contained peroxides capable of oxidizing iodide ion. Their choice of the ferroustitanous method over the iodometric methods was based on the greater sensitivity of the ferrous-titanous method compared with that of the Marks and Norrell method (4,with which high and variable blanks are obtained [the blanks by the modified Kokatnur and Jelling method (3, 7 ) are negligible]. Evidence has been obtained (6, 7 ) thpt the sodium iodideisopropyl alcohol method adapted from that of Kokatnur and Jelling (3) and the ferrous thiocyanate colorimetric method of Young, Vogt, and NeuTvland (8) are both basically accurate for the determination of peroxides in autoxidized materials which contain no conjugated diolefins. It was found that both methods give only empirical results on conjugated diolefin samples because of the slow rate a t which such peroxides are reduced and possibly because of side reaction. I n continuation of this study and in view of the wide use of the Yule and Wilson method, particularly in the petroleum industry, studies were made t o determine its fundamental accuracy in the determination of peroxides in autoxidized materials and to compare it with other methods. Since the correction factors and dilutions prescribed by Yule and m7ilEon (9) were admittedly arbitrary, they have not been applied in the work described in this paper. Specific peroxides of known purity, such as tetrahydronaphthyl hydroperoxide (tetralin peroxide), a,a-dimethylbenzyl hydroperoxide (cumene peroxide), teit-butyl hydroperoxide, benzoyl
peroxide, and hydrogen peroxide gave results ranging from 12 t a 80% of theoretical, with rather precise values for any given material. Results on autoxidized materials (nondienes) were generally from 20 to 60y0 of those by the sodium iodide method. Therefore, the evidence was strong that results by the ferroustitanous method are generally very low. Yule and Wilson ascribed this tendency to decomposition of peroxides in the presence of ferrous ion to form nonoxidizing products (such reactions with tetralin peroxide and other peroxides are well known), and stability of some peroxides toward reduction by ferrous ion. Yule and Wilson did not report the effect of molecular oxygen concentration on results by their method, but Lea (Q), employing a similar method, proved that values with air excluded were only a fraction of those obtained by the usual procedure performed in the presence of air. He concluded that atmospheric and dissolved oxygen was oxidizing ferrous ion in the presence of peroxide and that in the absence of peroxides such an oxidation was insignificant, The effect of lack of oxygen had previously been recognized by the present authors, who had drawn similar conclusions. However, in no cases were the values obtained kn0n.n to be higher than the theoretical ones. I n the light of later studies with the colorimetric method ( 2 ) which indicated that values obtained on autoxidized rubber samples were in no case higher than warranted by the amount of oxygen absorbed by the rubber, it seemed more likely that oxygen either catalyzes the reduction reaction or inhibits the ferrous ioncatalyzed decomposition reaction. Failure to obtain quantitative reduction of peroxides of known purity by the Yule-Wilson method even in the presence of oxygen is ascribed to too low a ratio of dissolved oxygen t o peroxide, this ratio being of the order of only one hundredth of that obtaining with the ferrous thiocyanate colorimetric method. The dependence of results upon sample size (peroxide concentration) is thought t o be closely related t o the oxygen effect. With the assumption that the relative results obtained by this method can be correlated with other properties of the material,
