Determination of Organic Peroxides by Reaction with Ferrous Iron I. %I. KOLTHOFF AND A. I. AIEDALIA' School of Chemistry, University of LMinnesota,Minneapolis, M i n n . Gross inaccuracies have been found in many published procedures for the determination of organic hydroperoxides by reaction with ferrous iron. In the light of recent work on the mechanism of this reaction, it appeared possible t h a t by proper choice of solvent and experimental conditions accurate methods could be developed. Two procedures have been studied, using aqueous acetone as the solvent in the absence of oxygen, and determining the extent of the reaction either by amperometric titration of ferrous iron with dichromate or by colorimetric determination of the ferric thiocyanate complex. Both procedures gave satisfactory results with cumene hydro-
T
HE determination of organic peroxides (particularly hydroperoxides) is ordinarii)- carried out by reaction with the reducing agents, ferrous or iodide ions. The principal advantage of the ferrous iron method is the greater sensitivity of detection of tlic reaction product (ferric iron determined colorimetrically as the t,hiocyanate). At the same time, the obvious sources of possible error in the iodometric method have frequently given rise to t.he belief that it is inherently less accurate than the ferrous method. Considerablr attention has been devoted to various sources of error in the iodometric method, including reaction of iodine with ethylenic linkages of the substratum, volatility of iodine, air oxidation of iodide, and slow reaction in the presence of traces of Ivater. Careful n-ork (2, 13, 23) has shown that suitable iodometric procedures give accurate results with pure peroxides; while in trace determinations (as with autosidized natural products) the m o r need be no grrater than 10%. 1
I'resent address, Brookha\-en National Laboratory. U p t o n ,
Table I.
L,I., 5 . Y .
peroxide and with autoxidized fatty acids; with methyl oleate hydroperoxide and autoxidized cottonseed oil, low peroxide values were found. The present work has demonstrated t h e significance of reaction mechanisms in the development of analytical procedures. Ferrous iron methods are inherently less accurate than iodometric methods for the determination of peroxides, so that, except in special cases, the iodometric methods are to be preferred for accurate work. For control work, where only relative peroxide values are desired, ferrous methods carried out in the presence of air are of particular value.
On the other hand, several-fold errors have been reported with various ferrous iron procedures. Table I summarizes the procedures that have been recommended; these differ from one another principally in the choice of solvent, exclusion of oxygen, and method of estimation of the extent of the reaction. In Table I1 are given the reported results obtained in the analysis of pure peroxides by these methods; Table I11 summarizes the results of intercomparisons of ferrous and iodometric procedures with peroxide-containing substrata. Taking the iodometric values as approximately correct ( 2 , is,23),it is clear that the ferrous method cannot be relied on for accurate results. Determinations carried out in the presence of oxygen have given results several times higher than the iodometrlc valucs. while in the absence of oxygen, values have been found which are several times too low. The procedures for which correct results have been reported are so closely similar to those for which groqs inaccuracies have been found, that the accuracy of all these procedures is questionable. In an attempt to clarify the qtatus of the ferrous iron method
Summary of Ferrous 1Iethods for Determination of Peroxides in Yarious Organic Substrata
(Determination carried out a t room temperature with air-saturated solutions, unless otherwise stated) Method of Detn. Solvent Acid Reagents (in Recommended of Amount of Additidn t o Ferrous Iron) Concn. of Active Iron Oxidized Subs t ra t u in and Concentrations Used; Oxygen, Na Gasoline Titration with 0.01 N T i C h Acetone-water (50-50), 0.2 N Less t h a n 0.013 S b sulfuric acid, thiocyanate Youn Vogt, and Nieuw- Gasoline Methanol. 0.03 jV sulfuric 2 x 10-6 t o 2 x Colorimetrically (as thiocyaianf'(~~) nate) 10-4 acid, thiocyanate Tanner and Brown ( 2 1 ) Gas oil, gasoline Titration with 0.02 .\' dichroLess t h a n 0.007.Vb Aoetic acid (93%) mate Lips, Chapman, and l l c - Fats a n d oils Acetone (96%), thiocyanate, 2 X 10-6 t o 3 X Colorimetrically (as thiocya10 -5 c Farlane ( 1 6 ) nate) no acid Lea (12) F a t s and oils Acetone-chloroform (88-12). Less than 0.0056 .Vb Titration with 0.01 .V TiCla 0.15 N sulfuric acid Bolland, Sundralingam, Rubber Benzene-methanol (72-28) 2.5 X 1 0 - 6 t o 5 X Colorimetrically (as thiocyaSutton, a n d Tristrain nate) 0.008 h' sulfuric acid, thio: 10 -5 cyanate (1) Robey and Wiew (EO) thiocya. Synthetic rubbers Chloroform benzene eth- 0 t o 2 x 10-4 Colorimetrically nate) anol (74-6-20) 0.04 N sulfuric acid, thidcyanate Laitinen and h-elson (11) Synthetic rubbers Benzene-methanol Colorimetrically (as ferrous o(91-9) Less t h a n 7 X lo-' phenanthroline complex) 0.028 M nitiic aoid, 0.0OOf N M DhosDhoric acld. 0.00035 M s u l f u h acid Authors and Reference Yule and Wilson ( 2 : )
-
a b e
-
Concentration in solution in which reaction between ferrous iron and peroxide takes place. Concentration of ferrous iron, Concentrations recommended by Lea (22) in a study of this method.
Notes Two-phase system
Nitrogen flushed (partially) 60-70' At 500
.it 500
ANALYTICAL CHEMISTRY
596 Table 11.
Comparison of Results Obtained by Several Methods of PeroxideAnalysis, w-ithPure or StandardizedPeroxides
Results ( % of Theoretical) Authors and Reference t o In Reference Peroxides Procedure I n air nitrogen Young, Vogt, and Kieuw- H202.succinyl peroxide Young et al. (26) Satisfactory accuracy ... land (86) tert-Butyl hydroperoxide, lauroyl peroxide Tanner and Brown ( d l 1 Tanner and Brown ( 2 1 ) 112, 133 67 HzOz, succinyl eroxide dihydroxyheptyl peroxide, Young et d. (26) Bolland et ~ l ( .I ) Satisfactory accuracy cyclohexene gydroperoxide Benzoyl peroxide, tert-butyl hydroperoxide Robey and Wiese (20) Robey and Wiese (80) Sati>factory accuracy Laitinen and Nelson ( 1 1 ) tert-Butyl hydroperoxide Robey and Wiese (20) u p t o 190 60,24 Kolthoff and Laitinen ( 8 ) tert-Butyl hydroperoxide Laitinen and Selson (1 I ) Up t o 690 118 Tetralin hydroperoxide, tert-butyl hydroperoxide, Bolland et QE. ( 1 ) Wagner et ai. ( 2 2 ) 85-115 4-63 cumene hydroperoxide, hydrogen peroxide .is above ( 2 2 ) Wagner et ~ l ( .8 4 ) Yule and Wilson ( 2 7 ) Values all less t h a n 100. Values lower depended markedly,oi than in air sample size
...
I
Table 111. Intercomparison of Results Obtained by Several Methods of Peroxide Analysis, with Peroxides Formed in Various Substrata Authors and Reference Young Vogt, and Xiedwland (96) Lips Chapman, and Rl'cFarlane (26) Lea ( l a ) Lea (12) Lea (18) Lea.(l2) Bolland et al. ( 1 ) Wagner et al. (22) Wagner et QZ. (24)
the primary (inducing) reaction in both cases being the stoichiometric ( 2 to 1) reaction between the peroxide and ferrous iron. Typical steps which are believed to take place are given below. Steps 1 and 2 form the stoichiometric reaction of a hydroperoxide with ferrous iion:
Ratio Found,
Substratum Gasoline
Reference t o Procedure a/b a. Yule and Wilson (27) (ferrous) 0.7 b. Young et QZ. (26) (ferrous) a . Lips et al. ( 1 6 ) (ierrous) 1.6-1-2.35 Fats and oils b. Lea ( 1 6 ) and Kokatnur and Jelling ( 6 ) (both iodometric) Olive oil, cod liver oil Q. Lips et al. (16) (ferrous) 2.16-2.58 h. Lea (14)(iodometric) Same a. Lea ( 1 % ) (ferrous) 2 . 4 2 - 3 66 b. Lea (14) (iodoinetric) Fats Q. Lips et al. ( 2 6 ) (ferrous) b. Q but with careful exclusion of air 3 . 3 -5.6 Same Q. Lips et al. ( 2 6 ) (ferrous) 5.4 -8.4 b. Lea (12) (ferrous, air excluded) 0.97-1.07 Rlethylcyclohexene, cyclo- a. Bolland et QZ. ( 1 ) (ferrous) hexene, dihydromyrcene h. Young et d. (26) (ferrous) Bolland et ~ l ( 2. ) (ferrous) 0.96-0.99 Cyclohexene, 2-pentene Q. b. Kokatnur and Jelling, modified (23) (iodoinetric) Bolland et QL. ( I ) but in absence of 0.17-0.22 Diisobutylene Q. air (ferrous) h. Kokatnur and Jelling, modified ( 2 3 ) (iodometric)
ItOOH RO.
+ Fe'+ +R O . + OH- + F e + + +
+ Fe+" f H + +
(1)
+ Fe+++
ROH
(2)
Competing with Step 2 is the reaction of the peroyide radical, RO., n-ith an organic compound, in Step 3:
RO.
+ R'H +ROH + R'*
(3)
examples of n-hich include the authors have carried out a fundamental study of the mechanism of the reaction between ferrous iron and peroxides under conditions of analytical importance (9). An extensive discussion of various proposed mechanisms of the reaction between ferrous iron and peroxides in the absence and presence of oxygen has been given elsewhere (18). The following scheme gives a satisfactory interpretation of the observed analytical results and is in accord with modern views concerning the nature of the peroxide-iron reaction. A peroxide molecule reacts with a single ferrous ion by a oneelectron transfer, in the course of which the peroxide molecule splits into a stable anion and a reactive free radical (Step 1 ) . The radical can undergo various subsequent reactions, the nature of which determines the over-all stoichiometry. Quantitative reaction of the radicals with ferrous iron (Equation 2) gives the ekpected over-all stoichiometry of two ferrous ions oxidized per molecule of peroxide reduced. Reaction of the radicals with organic compounds, in the absence of oxygen, leads to ferrous iron-peroxide reaction ratios of less than 2 t o l-i,e., to low analytical results. If oxygen is present, the (organic) radical may add on a molecule of oxygen, forming a peroxide free radical, reaction of which with ferrous iron leads ultimately to high analytical results Both the reactions which give high results and those nhich give low results are chain reactions, in which consumption of one radical always leads to the formation of another radical ; PO that the errors in either direction may be of considerable magnitude. Following classical terminology (IO), the low results may be ascribed to induced decomposition of the peroxide, and the high results, to induced oxygen oxidation of ferrous iron,
CH?.
CeI%C(CH,),
I
+ CsH5C(CHa)? I
0. CeH,C(CHa),
I
0.
----f
I
OOH (3a)
+ CHzCCHz + II 0 CsH,C(CH,), + CH3CCH2.k CHaC=CHr il
I
0
. OH
I
I
I + CeHjCCHa
OH
OOH
I
CsH,C(CH,)2
CeHbC(CH,),
(311)
0.
+ CEIaCHIOH +CeHsC(CH3)I + CHaCHOH
0.
OH
(3c)
The new radical, R'. , can reduce a ferric ion: CH,CHOH
+ F e + + +--+ CHICHO + F e + + + H +
(4)
or react with oxygen (if present), forming another peroxide molecule which can initiate reactions according to the above sequence:
+ +R'OO. + F e + + + H + --+ It'OOH + Fe+'+ R'.
R'OO.
0 2
(5)
(6)
The solvent can part,icipate in these radical reactions (Steps 3b, 3c). Radicals derived from the solvent are formed simply by transfer of a hydrogen atom from a solvent molecule to the radical derived from the peroxide. Radicals derived from organic solvents can undergo reactions analogous to those of the peroxide
597
V O L U M E 23, NO. 4, A P R I L 1 9 5 1 radical; however, different radicals d l be expected t o have different orders of preference for the various reactions which they can undergo. Under favorable circumstances, the over-all course of the reaction may be determined almost entirely bj7 the nature of the solvent radicals rather than of the peroxide radicals. Solvcnts irhich (in the absence of oxygen) give rise to low analytical results ma?- be termed “promoting” solvents, because they promote t h e induced deconiposition of the peroxide, while solvents w-liirh (in the absence of oxygen) lead to stoichiometrically correct analytical results may be termed “suppressing” solvents. In the reaction of ferrous iron with hydrogen peroxide, methanol and ethanol were found to belong to the promoting class; and acetone and acetic acid, to the partially suppressing (9). However, the behavior of a solvent in this regard will depend upon the particular peroxide, because the relative probability of various competing steps such as 3a, 3b, and 3c as compared to 4 will depend on the peroxide as well as on the solvent. .MI organic solvents which have been investigated give high results in the presence of oxygen. Some suppression of the induced oxygen oxidation of ferrous iron is brought about by chloride ion. Ileterminations carried out in promoting solvents in the presence of traces of oxygen may give nearly correct results, due to chance compensation of errors; indeed, incomplete removal of oxygen has actually hfaen recommended in some procedures ( 2 1 , 2.2). Similar compensation may be obtaincti in air-saturated solutions nith rclativcly high concentrations of peroxide (9). However, this compensation of gross errors evidently cannot serve as the liasis for a reliable method of peroxide determination.
Table IV. Determination of Hydrogen Peroxide and Cumene Hydroperoxide by Reaction with Ferrous Iron (0.002M), in Aqueous Solution, in Absence of Oxygen Peroxide
Other Compounds
Molar Reaction Ratio
Correct results can be obtained if oxygen is completely absent and if a solvent is used which entirely suppresses the induced decomposition of the peroxide. In view of the above discussion, this appear5 to be the only sound basis for the development of satisfactory procedures. Illustrative data obtained in connection n-ith the fundamental studies (9) are given in Table IT’. Pnder the conditions of these experiments reasoniibly accurate tleterniinations were made of hydrogen peroxide in the prcsence uf a small amount of ethanol, and likewise of cuniene hydroperosidc, using acetone, acetic acid, or chloride ion as suppressors. Icon-ever, the technique employed was not convenient from an tinalj-tical standpoint, because the solutions were subjected to prolonged flushing with nitrogen in special vessels, and the excess of ferrous iron was titrated with ceric sulfate, requiring the addition of considerable water in order to obtain good end points in the experiments with acetone or acetic acid as solvents. The follon-ing work has therefore been carried out with the objective of developing simple, fundamentally sound analytical procedures for the accurate determination of organic peroxides, particularly peroxides present as traces in autoxidized substrata. MATERIALS AND TECHNIQUES
Acetonc. is an excellent solvent for a num1)er of important natural products, including fatty acids, fats, and many prtroleum products; for the determination of peroxides in these substrata, acetone is particularly suitable as a solvent, inasmuch as after
renioval of oxygen-13-liich is readily brought about by boiling for a short time a t a relativrly low tmiperature-it tends to suppress the induced decomposition of the peroxides. Acetone has therefore been used as a solvent in a11 the present work. Ordinary C.P. acetone, as received in drum lots, contains significant amounts of perouides. A single distillation through a short unpacked column, a t a rate of about 1 liter per hour, gives a product of adequate purity for use in the titration method, described below; however, for use in the more sensitive colorimetric method, it was found necessary to add ferrous iron to the first distillate and then redistill. To 4 liters of freshly distilled acetone are added 10 ml. of 0.5 J-1 ferrous perchlorate in 0.4 -V perchloric acid (prepared as below); then the mixture is redistilled a t a rate of about 1 liter per hour. The redistilled acetone may be kept for several days in a full bottle. The blank due to the aretone should correspond to less than 3 X 10-7 .li ferric iron. Ferrous sulfate, although sufficiently soluble in 80% acetone t o permit its use in the titration procedure, is extremely insoluble in acetone containing only a few per cent of water, and is therefore not suitable for use in the colorimetric procedure described below. For use in this procedure, solutions of ferrous perchlorate in perchloric acid have been prepared by either of the following techniques, which were found to give a minimum of air oxidation of the ferrous iron during preparation. 1. Preparation of 0.5 M Ferrous Perchlorate in 0.4 N Perchloric Acid. Place 175 ml. of 207, perchloric acid (Baker’s) in a 500-ml. flask with a reflux condenser and ground joints. Boil for 10 minutes while passing nitrogen through the solution. Add 7.00 grams of pure iron wire and continue to boil, under nitrogen, until the iron is dissolved (about 2 hours). Cool under nitrogen and make up to 250 ml. with water. Transfer 50 ml. to a loosely stoppered 2-ounce bottlc, add 0.05 gram (6 inches) of iron mire, and keep a t 50” overnight to reduce traces of ferric iron; then cool under nitrogen and keep under nitrogen a t room temperature. Small pieces of iron wire, from the disintegration of the &inch length of wire, may be suspended, but are not harmful in the peroxide determination. 2. Preparation of 0.1 MFerrous Perchlorate in 2 MPerchloric Acid. Take 0.28 gram of iron wire and 50 ml. of 207, perchloric acid in a loosely stoppered 2-ounce bottle; keep a t 50” overnight to dissolve; cool, and store under nitrogen. I n solutions prepared according to the above procedui e-, the ferric iron content should be less than 0.1 or 0.47,, respectively, of the ferrous iron content. Other inorganic reagents used were of C.P. or analytical reagpnt grade. hlerck reagent grade ammonium thiocyanate has been used without further purification; solutions of 11 grams of ammonium thiocyanate in 100 ml. of water or aretone are referred to below as “14% solutions.” The following peroxides were used in the present study: hydrogen peroxide, Merck’s Superouol, 30%, C.P. Cumene Hydroperoxide (CHP). A sample furnished by the Hercules Powder Co. was purified by the method of Hock and Lang ( 4 ) ; analysis by an iodometric macroprocedure indicated a purity of 94.1%. Methyl Oleate Hydroperoxide (MOHP). The sample was prer>ared bv and received from Lundbera of the Hormel Institute of the Uni;ersity of Minnesot,a; iodonietric analysis by the procedure of Kokatnur and Jelling (6) indicated a purity of ss%. Fatty acid prepared from commercial soap called S.F. flnkcs mas obtained from the Procter I% Gamble Co. The pFrouide content of this fatty acid mixture was increased by passing air through the melted acid a t 70” C. for 24 to 48 hours. Cottonseed oil, manufactured by the Southern Cotton Oil Co., was used; peroxides were formed by passing air through a t 70” for 20 to 30 hours.
Tn-o general methods of analysis have been studied, differing chiefly in the composition of the solvent and the method of determining the extent of the perouide-ferrous iron reaction. The titration procedure, which is carried out in 70 to 80% aqueous acetone, is particularly convenient for soaps and fatty acids, while the colorimetric method, carried out in 06 to 98% acetone, is applicable to fats and oils. Analysis of petroleum products has not been studied in the present JTork; however, these products are in general soluble in 06 to 987, acetone.
ANALYTICAL CHEMISTRY
598 Table V. Peroxide Values Obtained after Various Times of Standing (Titration in presence of sir) Reaction Time, Sample Min.
S.F. flakes (Procter 8: Gamble soap)
30
60 120 10
CHP
30 120
“Apparent” Active Oxygen, P.P.31.
290 290 320 269,000 272,000 271,000
TITRATION METHOD
General Considerations. I n this method, an excess of ferrous iron is added to a solution of t.he peroxide in aqueous acetone; after reaction, the excess of ferrous iron is determined by amperometric titration with dichromate. Determination of the extent of reaction in this manner is inherently capable of great precision, and has the advantage over colorimetric methods that the composition of the solvent need not be fixed.
=t- \ \ \ 4
Figure 1. Titration Curves of Ferrous Iron in Aqueous Acetone with 0.00500 N Dichromate Using rotating platinum microelectrode a t $0.72 volt C8. S.C.E. 1 . 2 . 3 . Titration of excess ferrous iron after reduction of peroxides i n Emory single-pressed stearic acid 4, 5. Standardization of ferrous iron solutions
Amperometric Titration of Ferrous Iron in 70 to 80% Acetone. The titration is carried out using a rotating platinum microelectrode a t a potential of +0.70 volt with respect to the saturated calomel electrode (S.C.E.). The current before the end point is an anodic one owing to oxidation of the ferrous iron a t the platinum wire. This current is measured on a microammeter or sensitive galvanometer and is plotted against the amount of dichromate solution added. The current is zero a t and after the end point. The end point can be estimated graphically to 0.02 or 0.03 ml. of 0.00500 N dichromate. After several titrations, the current from a given excess of ferrous iron usually becomes smaller and the end point cannot be estimated as accurately. Heating the platinum electrode to redness for a few seconds with a fine gas-air flame from a hand torch restores the current to its original high value; if this is done carefully, there is no danger of cracking the glass tube to which the electrode is sealed. The electrode should be of the shape recommended by Kolthoff and Harris ( 7 ) . Titration Procedure in Presence of Air. An appropriate amount of soap soap solution (containing not more than 10 ml. of water), fatty h i d , or other peroxide-containing substance is dissolved in 100 ml. of acetone and acidified with 10 ml. of 1 to 5 sulfuric acid, and 10 ml. of ferrous ammonium sulfate solution, approximately 0.007 M in 0.1 N sulfuric acid, are added with stirrmg. The solution is allowed to stand in a’250-ml. beaker for 30 minutes, open to the air. The excess of ferrous iron is back-
titrated with 0.00500 N dichromate by the amperometric procedure described above. If the excess of ferrous iron is less than 1 ml. (as can easily be estimated from the initial galvanometer reading), another 10-ml. portion of ferrous solution should be added and the mixture allowed to stand another 30 minutes before titration. The titer of the ferrous iron must be determined daily. Calculation. The active oxygen content of the sample is calculated as follows. Subtract the number of milliliters of dichromate used in titrating the sample from the number of milliliters of dichromate used in the blank. Let this difference be D. Then P.p.m. of active oxygen =
(4O)D grams of sample
I t is assumed that those oxygen atoms in the peroxide which are “active” have a valence of 2, or an equivalent weight of 8. Hydroperoxides are reduced to alcohols, so that only one oxygen atom of the two oxygen atoms in the hydroperoxide group is active; thus the total oxygen in one hydroperoxide group is twice the amount of “active oxygen.” Because reduction procedures give a direct estimation of active oxygen, it is desirable to express results in terms of active oxygen. Development of Titration Procedure. Determination of the extent of the reaction was first attempted by titration of the excess ferrous iron with ceric sulfate, using ferrous phenanthroline indicator. However, the end points observed in 70% acetone were not sharp, a precision of only about 0.2 ml. of 0.005 N reagent being obtainable. The much greater precision of amperometric titration is illustrated in the typical titration curves of Figure 1. The amperometric titration is a modification of one previously developed in this laboratory (17). I t 11-asreported that in aqueous acid a well defined anodic diffusion current is obtained for ferrous iron (10-5 to 10-4Jf) with the rotating platinum microelectrode, in the range +1.0 to $1.2 volts (us. S.C.E.) Attempts to reproduce this result in the present study have been unsuccessful; some of the polarogranis obtained are shown in Figure 2. A4 polarogram of 0.0012 Jf ferrous ammonium sulfate in 70% acetone, 0.5 N in sulfuric acid, is given in Figure 3. Here also a good diffusion current is not obtained. The wave is shifted by about -0.4 volt when acetone is used rather than water; nevertheless the current obtained in aqueous acetone a t a potential of $0.70 volt is fairly constant and is proportional to the amount of ferrous iron, as shown by the titration curves, so that the amperometric titration is satisfactory even though a good diffusion current is not obtained. Separate experiments have established that the amperometric titration of ferrous iron gives results which are accurate as well as precise. Results Obtained in Presence of Air. Although the titration itself is accurate, the reaction between the peroxide and ferrous iron is not stoichiometric in the presence of air; as shown below,
I
I
I
I
I
POTENTIAL OF ROTATING ELECTRODE (VOLTS VS. S.CE.1
Figure 2.
Current-Voltage Curves
Obtaine 1 with rotating platinum microelectrode i n aqueous solutions of ferrous iron. All solutions airsaturated. Temperature, 30‘ C. Electrode, platin u m wire 0.7 mm. diameter, 3 mm. long, sealed i n side of 6-mm. glass tubing
1. 0.1 2. 1.1 3. 2.2 4. 4.4 5. 8.8
M X X X X
HClOi (residual current) 10-6 M Fe(ClO4)p i n 0.1 M HCIO4 10-5 M Fe(ClO4)r in 0.1 M HClO4 10-5 M Fe(ClO4)z in 0.1 M HC104 10-5 M Fe(C1Oi)z in 0.1 M HClOd
V O L U M E 23, NO. 4, A P R I L 1 9 5 1
599
considerable induced air oxidation of ferrous iron takes place. Kevertheless, if the extent of this induced reaction is reproducible, results of peroxide determinations will be reproducible, so that the method may have practical value. With cumene hydroperoxide, agreement of duplicate analyses is within 1 to 2% in the presence of air. With soap peroxides, agreement has been found in the range 5 to 10% in the presence of air. With cumene hydroperoxide, identical results were found after 10, 30, and 120 minutes' reaction time; with soap peroxides, a slow increase in apparent active oxygen content was found after different times of standing, as shown in Table V. Probably the peroxides present in commercial soaps or fatty acids are of several types; it is reported by Wagner et al. ( 2 2 , 2 4 )that bridge-type peroxides, which are formed from fatty acids containing conjugated double bonds, are much less reactive toward ferrous iron than are hydroperoxides. The time of 30 minutes, recommended in the procedure given above, appears to be sufficient for reaction of nearly all the peroxides present in the samples studied. Experiments with cumene hydroperoxide have shown that the results in the presence of air are independent of sample size or of amount of ferrous iron present,provided the latter is inat least 25% excess. Passage of air, or even of oxygen, through a reaction mixture of ferrous iron with a peroxide-containing fatty acid gave results nearly identical with those obtained by the above procedure, in which the reaction mixture was left open to the air. It can be calculated from the solubility of oxygen in acetone that the concentration of oxygen in air-saturated reaction mixtures is well in excess of that consumed in the observed induced reaction. Table VI. Comparison of Ferrous Procedure (in Air) with an Iodometric Method Expt. No.
Description of Sample
1 Emery single-pressed stearic acid 2 Same, air passed through melted f a t t y acid a t 70° C. for 50 hours 3 Emery single-pressed stearic acid, heated 6 hours in
Active Oxygen, P ,P. 1LI.a IodoFerrous metric 220 108 310
185
oven a t 100° C . 30 14 4 S.F. flakes 6 12 5 F a t t y acid from (4) oxidized by passing air through for 15 hours a t 70' C. 2140 525 G Armour soap flakes 2 10 7 F a t t y arid from (e), air passed through melted f a t t y acid a t 70' C. for 20 hours 400 94 8 F a t t y acid from (61, air passed through melted fatty 3080 1470 acid a t 70" C . for 48 hours 9 S.F. flakes, allowed t o react with persulfate in aqueous solution for 12 hours a t 50' C. in presence of air 620 270 10 Soap from Armour Neo F a t S o . 3R (about 50% linoleate), aqueous solution exposed t o air a t 50' C. for 12 hours 750 330 11 Pure stearic acidb, heated 6 hours a t 100' in air 660 180 12 Pure palmitic acidb, air passed through melted f a t t y acid a t 70" for 6 hours 310 110 a Based on dry soap or fatty acid. b Prepared by K. 11. Lauer, Organic Chemistry Division, University of hlinnesota.
Comparison of Ferrous and Iodometric Methods. The active oxygen content of a number of samples of peroxide-containing soaps and f a t t y acids has been determined by both the ferrous method in the presence of air, as given above, and the iodometric procedure of Kokatnur and Jelling (6). About 0.5 gram of the f a t t y acid as such, or as precipitated from a soap solution, was taken for the iodometric determination, Some typical results obtained with various soaps and fatty acids are given in Table VI. The values obtained by the ferrous method (when not negligibly small) are always much higher than those obtained iodometrically, the ratio being in the range 2 to 5 . It has been shown ( 1 )that in the reaction between ferrous iron and cumene hydroperoxide in aqueous solution a similar effect is caused by induced air oxidation of the ferrous iron. T h a t the present results are to be interpreted in the same manner is shown by the data of the following section, in which active oxygen contents lower than the iodometric values were found with cumene
hydroperoxide by the ferrous method in the absence of oxygen. The factors influencing the magnitude of the induction factor in the determination of soap peroxides in the presence of air have not been studied. It appears possible that the induction factor may depend upon the nature of the particular peroxides, and upon the composition of the fatty acids. Titration Procedure in Absence of Oxygen. As shown below, the following technique gives satisfactory removal of oxygen from the reaction mixture.
z
25
-
to
-
2
5 I5
-
Y
(I
!?I
2 IO 0 -
o
l .D
/
I ~
O
J
I O 4 0 5 0
I
I
l
m . 7 o a o
POTENTIAL OF ROTATING ELECTRODE (VOLTS VS. S.C.E.)
Figure 3.
Current-Voltage Curve of 0.0012 M Fer-
rous Ammonium Sulfate in 70Yo Acetone, 0.5 A' in
SulfuricAcid, with Rotating Pla tinum Microelectrode
A suitable amount of sample (depending on the peroxide content) is weighed into a 250-ml. Erlenmeyer flask. Approximately 150 ml. of distilled acetone are added and the flask is placed on a hot plate. A few glass beads are added to prevent bumping. The solution is allowed to boil rather vigorously for 15 minutes before the other reagents are added. Meanwhile purified nitrogen is passed through an 0.01 LV solution of ferrous ammonium sulfate in aqueous 0.1 M sulfuric acid and through a 6 N solution of sulfuric acid, a t room temperature, for a t least 45 minutes. Approximately 10 ml. of sulfuric acid are added slowly from a nitrogen-filled pipet to the boiling acetone, then exactly 10 ml. of ferrous solution are added similarly. The reagents must be added so slowly that boiling never stops and no air is drawn into the flask. The solution is kept boiling for 15 minutes longer, and then it is poured into a 250-ml. beaker. The flask is rinsed with distilled acetone and the rinsings collected in the beaker. The solution titrated amperometrically with O.qO500 N potassium dichromate, using a rotating platinum mcroelectrode a t a potential of +0.70 volt with respect to the saturated calomel electrode as described previously. A blank is run in the same manner. The calculation is performed as described previously. Results Obtained in Absence of Oxygen. Data obtained in the absence of oxygen, by the above procedure with various modifications, me given in Tables VI1 and VIII. The data of Table VI1 were obtained with samples of S.F. fatty acid through which air had been blown a t 70" C. for different lengths of time (several days). The data of Table VI11 were obtained with purified cumene hydroperoxide and other peroxides.
Table VII. Determination of Active Oxygen Contents of S.F. Fatty Acids by Titration Method in Absence of Oxygen Method of Treatfnent before a n d d u n n g Reaction Boiled b
Time of Treatmenta, Min. 15
Time of Reaction, Min. 10 15
Active Oxygen,
P.P.M.
Ferrous
Iodometria
540 575 600 585 580 1090
640
640 640 640 640 30 BoiledC 15 1140 Boiledb 30 1140 1 hour 980 Nitrogen, room temp.b 1 hour 1110 1140 1 hour a Before adding sulfuric acid a n d ferrous solution (nitrogen-saturated). b Container, 250-ml. Erlenmeyer flask. Initial volume of acetone, 150 rnl. 0 Container, 500-ml. Erlenmeyer flask. Initial volume of acetone, 300 ml. 15 15 15
20
30 15 10
ANALYTICAL CHEMISTRY
600 Table VIII. Determination of Active Oxygen Contents of Peroxides by Recommended Titration Procedure in Absence of Oxygen Peroxide or Active Oxygen, P.P.M. Substratum Ferrous Iodometrii 93,600 99,300 CHP 8,600 43,400 MOHPa 170 855 Cottonseed oilb a Methyl oleate hydroperoxide. b Reaction mixture 90% in acetone.
I t is seen from Table VI1 that the recommended procedure (line 2) gives peroxide values which are slightly lower than the values obtained iodometrically. The low results are not due to decomposition of the soap peroxide during boiling, as the same results were found in experiments in which the peroxide-acetone solutions were boiled for 15 and 30 minutes. The data of Table VI1 show that 15 minutes' boiling gives satisfactory removal of oxygen, and 15 minutes' reaction time is sufficient at the boiling temperature. The method involving boiling of the solution is recommended in preference to that involving removal of oxygen by means of flushing with nitrogen, because the former method is much more rapid. In cases where the peroxides are very unstable, however, i t would be preferable to flush u ith nitrogen a t or below room temperature. K O reaction between ferrous iron and oxygen takes place after the reduction of the peroxide, either upon exposure to air during the transfer of the solution from the flask to the beaker, or during the titration in the presence of air. Although the titration procedure in the absence of oxygen gives results of the correct order of magnitude, with fatty acid peroxides or with cumene hydroperoxide, it is seen from Table VI11 that very low results are obtained with methyl oleate hydroperoxide and cottonseed oil peroxides. Discussion of this effect is deferred until after consideration of the colorimetric method.
excess of ferrous iron over peroxide must be used; ~ i t h i nthe range studied, it was found that as the water content of the system was decreased, a large excess of ferrous iron was required to obtain the stoichiometric reaction with cumene hydroperoxide. Too large an excess of ferrous iron is undesirable from the standpoint of the correction for the blank determinations. To keep the substratum (fat, etc.) in solution, the xatrr content must be kept below 2 to 4%. T o obtain an accurate colorimetric determination of the ferric iron formed, the water content must be known within a few tenths of 1%; therefore, the loss of water during boiling must be small. T o prevent decomposition of the peroxide during the boiling which precedes the addition of ferrous iron, the solution must not be strongly acidic during this stage. To prevent solvolysis of the ferric iron formed, the solution must be strongly acidic after addition of the ferrous iron. The amount of acid required to prevent solvolysis of the ferric iron, as well as the extent of decomposition of the peroxide in the presence of a given amount of acid, depends upon the water content of the solution. Determination of Ferric Iron. I n the procedure as outlined above, ferric iron is formed by the peroxide-ferrous iron reaction and is determined colorimetrically with thiocyanate. The variables affecting this determination have been studied systematically; the results are given in Tables I X and X. In all determinations a Leitz-Rouy photrometer was used, with a l-cm. square cell; the filter which gave minimum per cent transmittancy was that at 460 mp. Table IX. Effect of Various Amounts of Water and Acid upon Intensity of Color with a Given Amount of Ferric Iron [All solutions were 2.30 X 10-5 M in Fe(ClO4)s and 0.28% in S H I C S S ] Water, %
HClOJ, .M X
% Transmittancy
4.8
after 5 XIinutes 36.3 37.5 40.0 37.9 38.2
20.8
40.0
lo3
2.8 2.8 2.8 1.8
COLORIMETRIC METHOD
General Considerations. The following work was carried out q i t h the objective of developing a procedure applicable to a wide range of substrata. Acetone, the use of which is particularly advantageous from the standpoint of suppressing the induced reactions, is also an excellent solvent for many natural products, ii not more than a small amount of water is present. The color reagent selected was thiocyanate, the use of which in acetone-water systems has been described in the literature ( 2 5 ) ; this reagent permits the direct estimation of the amount of ferric iron formed, and the sensitivity of the test is particularly high in a solvent of low dielectric strength, such as an acetonewater mixture of high acetone content. In contrast to most other color reagents for iron, the thiocyanate procedure is satisfactory in strong acid, so that the problem of selecting buffers for use in 96% acetone does not arise. The following general procedure was adopted as a basis for study. Procedure. Dissolve the sample in about 80 ml. of acetone in a 100-ml. volumetric flask; add a few glass beads and heat to boiling. Continue boiling for 5 minutes; then add a solution of ferrous iron in acid (kept air-free and of a minimum volume). Continue boiling for 5 minutes; cool, add acid if necessary, add thiocyanate, make up to volume with acetone, and read after 5 minutes. In the development of a satisfactory method, the following factors must be considered and, in many cases, balanced against each other. Not all factors were, of course, evident a t the outset of this work. The solution must be free from oxygen before the addition of ferrous iron and must be kept oxygen-free until the peroxide is completely reduced. Both ferrous and ferric iron must be present in a soluble form. T o suppress the induced decomposition of the peroxide, a hrge
Table X . Solvolysis of Ferric Iron on Boiling in Acetone Solutions Concentrations (Based on 100 511. of Solution) Fer++ HC104, Water, ,vi x io; M x 103 % 2.50 2.8 2 2,50 2.8 2 2.50 2.8 2 2.50 2.8 2 2.50 0.8 0.3 2.50 0.8 0.3 2.60 20 2 2.50 20 2 2.50 10 2 2.50 10 1 2.50 10 0.5 2.50 5 2
T~~~ of Boiling, Min.
..
5
Transmittancy,
. %
40.8
From the above data it is seen that within the range studied, the intensity of the color increases somewhat as the water content is decreased; this agrees with the trend established by Woods and hlellon (86), who worked in the range from 0 to 80% acetone. Relatively little drpendence upon the acid concentration is found. Because in the procedure outlined above the solution is boiled and then cooled before adding thiocyanate, it seemed advisable to check the effect of temperature upon the intensity of the color developed. This is of particular importance in the determination of fats, as it was found that with a 0.4% solution of a typical fat (Crisco) in 98% acetone containing 0.25% of ammonium thiocyanate i t was necessary to keep the temperature above 30" C. in order to prevent precipitation of the fat. In a brief study of the effect of temperature and time of standing upon the intensity of color developed according to the above procedure, i t was established that sufficient accuracy (l'%)can be obtained nithout pre-
V O L U M E 2 3 , NO, 4, A P R I L 1 9 5 1 cise control of temperature; the solutions (after boiling) may l)c cooled to between 15" and 38", thiocyanate added, and the color moasured within a few minutes ( 3 to 7 minutes, at 15a to 25 ', or 2 t o 4 minutes a t 38"). Fading of the ferric thiocyanate color, which is more rapid a t 38" than at 25", is so rapid in the tioiling solution that if the thiocyanate is added to the boiling solution before cooling, the first readings of transmittancy are too high. The experiments of Table X were carried out in order to estal )lish the concentration of acid required to repress the solvolysis of ferric iron in acetone-water mixtures of varying composition. R(:causo the solvolysis is much more rapid in the boiling solution than a t room t,emperature, solutions of f m i c perchlorate and perchloric acid in aqueous acetone (approximately 80 nil.) were tjoiled for various lengths of time, then cooled to room teniperature, treated with thiocyanate, and made up to volume (100 ml.) with acetone; a a t e r was added a t this stage if necessary eo that all solutions should contain t,he same concentrations of thiocyanate, water, and acid a t the time of reading. I t is seen that with 2 ml. of water, suffioient perchloric acid must tie addfld to make the h a 1 solution 0.02 -21,in order that the solvolysis of ferric iron may not be serious during 5 minutes of hoiling. With 0.3 ml. of water, the solution nerd Ije only 0.0008 AI in prrchloric acid. Thus, as expected with this very strong acid, a givcri amount of perchloric acid is more effective in suppressing if the solution contains li.ss water. Construction of a Calibration Curve. A straight-line (scmilog) curve of per cent transmittancy us. concentration of ferric iron was constructed from experiments carried out as follow : .In accurately weighed amount of pure iron wire was dissolved in perchloric arid, then oxidized with an excess of hydrogen peroxide, which was destroyed by boiling for several minutes. This solution of ferric perchlorate (0.02505 M ) a-as diluted 1 t o 10 in acetone, 0.04 JI in perchloric acid. Portions of this solution w r e mixed in aoetone with thiocyanate, water, and perchloric acid, to give solutions which were 0.2870, 2Y0, and 0.001 JI in these ingredients, respectively. After standing 5 minutes a t room temperature, the per cent transmittancy was read a t 460 nip in the Leitz-Rouy photrometer. .4 solution 6.0 X 10-j M in ferric irou gave 10% transmitt,ancy. Decomposition of Cumene Hydroperoxide in Acid Solution. .is shown atiove, a fairly high acid strength is necessary to repress solvolysis oi ferric iron. However, it is known (-5) that the decomposition of hydroperoxides is catalyzed by acid. The data of Table X show that, if a small amount of perchloric acid is added licifore the cumene hydroperoxide solution is boiled to wniove oxygen, nearly complete decomposition of the cumene hydropcrosidc orcurs during this boiling. If no acid is added, a small amount of decomposition takes place, probably as a result of traces of acid in the acetone. -4ddition of sodium hydroxide or quinoline prevents this decomposition. -4ddition of a small amount of n-ater (in the absence of added acid) also prevents this decomposition, owing to decrease in the effectivc strength of the acid impurities due to change of the solvent (cf. Table SI). Aiccordingly,in the procedures whirh have becn developed for. the tletrrmination of peroxides, a small amount of water is added before boiling. Of course, it is necessary to add acid :it thr time of addition of ferrous iron, both because the prxroxidr-ferrous iron roaction consumes acid, and to prevent the solvolysis of fcrric iron. In the acid solution containing ferrous iron, competition occurs twtn-eon thc acid-catalyzed decomposition of eumene hydropcrosidc and the ferrous iron-cuniene hydroperoxide reaction. The fact that stoichiometric results are obtained with cumenc hydroporoside according to the procedures given ticlow is evidence that, a t least with this particular peroxide and under the experimental conditions, the reaction with ferrous iron is much mor(' rapid than the acid-catalyzed decomposition. Considcration was given t o adding together with the ferrous iron only sufficirnt acid to participate in the ferrous iron-peroxide reaction, and thon, after this reaction is complete, adding a relatively l u g e :imoulit of prrchloric acid to dissolve the solvolyzed fcrric iron.
601 However, preliminary experiments indicated that traces of oxidizing agents were present in the perchloric acid, so that a large blank coirection would have to be made; for this reason i t seemed preferable to add all the acid in a single solution with the ferrou? iron, reduction of the oxidizing impurit,ies in the acid having been effected during solution of the iron wire in the acid.
Table XI.
Decomposition of Cumene Hydroperoxide during Boiling Found
~~
\\.ater, n /c
...
... 0.2 0.1 0. 1 0.1 0.2 0.4
Initial Conrentrations CHP, .M 106 Other
x
0.392 1.18 1 . I8 1 . 21 1.21 1.21 1.21 1.21
........ ........
I-IClOa, 1 X 10-5 M NaOH, 1 X 1 0 - 5 M Quinoline, 1 X 10-5 .If
., .,.. . ...... . ........ . .
CHP-
Fe+++ .l;i X Ib3
(70 of C H P
0.73 2.36 0 , o . 11 2.39 2.37 2.40 2.40 2.42
93 92 0, 5 99 98 99 99 100
taken)
The results of Tahlc S I were obtained with solutions of purified cumene hydroperoxide in approximately 80 ml. of acetone, to which were added various ingredients, in concentrations based on t,he final volume of solution (100 ml.); the solutions were boiled 5 minutes, and then ferrous perchlorate was added. After furt.her boiling (5 minutes) the solutions were cooled, treated with thiocyanate, and analyzed photometrically. The concentrations of cuniene hydroperoxide taken, in the second column of Table S I , are based upon the amount of cumene hydroperoxide weighed out for the stock solut,ion, corrected for the purity of this samplc (94.1%). The concentrations of ferric iron (fourth column j were cstimated from the photomet,er readings hy use of the calibration curve, and corrected for the small blank. The amounts of cumene hydroperoxide found (last column) w e expressed in per cent of the amounts of cumene hydroperoxide taken, and were calculated from the coneentrat,ions of ferric iron on the assumption that the ferrous iron-cumene hydroperoxide reaction is stoichiometric (reaction ratio 2 to 1). Determination of Cumene Hydroperoxide. The determination of cuniene hydroperoxide h m been st,udicd according to the general proccdure given above, with the follonhg specific modifications. I n Procedure 1, no water is added t o the solution until the ferrous iron is added. The iron is added as a nitrogen-saturated aqueous solution, 0.5 .If in ferrous perchlorate and 0.4 M in perchloric acid; 0.2 ml. of this solution is added from a graduated 1-nil. or 0.2-ml. pipet. After boiling and cooling, 0.2 ml. of 0.5 .If pcrchloric arid is added, then 2 ml. of 14% ammonium thiocyanate (aqurous solution). I n Procedure 2, 0.2 ml. of water is added before the solut,ion is heated. The procedure is otherwise identical with Procedure 1, rTxcept that ( t o keep the total volume of water approximately the same j 0.05 ml. of 2 111 perchlorir acid is added after boiling and cooling. I n Procedure 3, 1 . i 5 nil. of water are added before heat,ing; 0.25 ml. of a 0.1 Jf solution of ferrous perchlorate in 2 iM perc-hloric acid is added after 5 minutes' boiling; after 5 minutes' further boiling, the solution is cooled and 2 ml. of a 14% solution of ammonium thiocyanate in acetone are added, and the color is read as usual. Procedure 1 \vas studied before the data of Table XI were obtained. In view of these results, it is to be expected that the highest values of cuniene hydroperoxide found, according to Procedurc 1, would be ahout 92% of the cuniene hydroperoxide taken (owing to decomposition of about 870 of the cumene hydroperoxide before addition of the ferrous iron). This result y a s actually found; over the range of concenhtions of cumene hydroperoxide from 0.37 X 10-6 JI'to 2.91 X .M,the amount of cumene hydroperoxide found was 89 to 0270 of that taken. That this is not due to incomplete reaction was established by boiling for 10 minutes (instead of 5 ) after adding ferrous iron; again a value of 92y0 was found.
ANALYTICAL CHEMISTRY
602 With Procedure 2, the amounts of cumene hydroperoxide found were 98 to 100% of that taken, over the range 0.60 X 10-6 M to 2.42 X 10-6 M ; thus, suppression of the acid-catalyzed decomposition of cumene hydroperoxide, as well as of the induced decomposition of cumene hydroperoxide, is achieved under the conditions of this procedure. I n both Procedures 1 and 2, use of one fourth the stated amount offerrousiron-i.e.,2.5 X Minstead of 1 X lO-3M-gave low results (78 and 86%, respectively, due to incomplete suppression of the induced reaction. With Procedure 3, on the other hand, good results were obtained with this amount of ferrous iron (Table X I ) . It appears that suppression is enhanced by the presence of a small amount (2%) of water. In the amperometric titration technique described above the solvent contained 20 to 30% of water, and satisfactory results (with cumene hydroperoxide) were obtained with only a twofold excess of ferrous iron.
Table XII.
Determination of Cumene Hydroperoxide According to Procedure 3
(All concentrations based on 100 ml. of solution) Concentration5 Found C H P Taken, M X lo6 F e + + + ,M X 106 C H P , % of C H P taken 0.86 0.42 102 1.75 0.85 103 3.48 103 1.69 5.00 99 2.52 8.50 338 1 . 26a a Ferrous iron added as soon as solution started to boil.
The last experiment of Table XI1 illustrates the large error resulting from incomplete removal of oxygen from the acetone solution, before addition of the ferrous iron (cf. Table VI). Further data establishing the time of boiling required before addition of the ferrous iron, in order completely to remove oxygen from the solution, were obtained with a sample of oxidized S.F. fatty acid (Table XIII). A time of 3.5 minutes is seen to be nearly adequate, but 5 minutes is preferable. Longer boiling before adding the ferrous iron does no harm; but boiling for 10
Table XIII. Removal of Air from Reaction Mixtures after Various Times of Boiling [All solutions contained 29.7 mg. (per 100 ml.) of oxidized S.F. f a t t y acid (iodometnc:950 p.p.m.). Procedure 3 followed throughout] Time of Boiling, Minutes Fe + Found, Before Fe + After Fe ‘M x 10s + +
+
+ +
5 5
1
2
3.5 5 10 5
10
9.20 5 2fi
2.63 2.49 2.49 2.38
Table XIV. Determination of Peroxides According to Procedure 3 Ratio of Ferrous t o Found, Active Oxygen Iodometric nig./ioo MI. M x 105 Found, P.P.M. Active Oxygen A. Methyl oleate hydroperoxide (iodometric. 43,400 p.p.m.) 1.24 20,300 0.489 0.47 0.978 0.41 2.17 17,800 2.93 5.86 16,000 0.37 B. Oxidized f a t t y acid (iodometric, 950 p.p.m.) 5.94 0.62 830 0.87 14.9 1.33 715 0.75 29.7 2.49 670 0.71 59.4 5.24 705 0.74 C. Oxidized cottonseed oil (iodometric, 620 p.p.m.) 35.7 0.49 110 0.18 71.4 1.02 114 0.18 125 1.75 112 0.18 179 2.84 127 0.20 267 3.81 114 0.18 357 4.97 111 0.18
Sample Taken,
Fe
+ + +
Table XV. Results in Presence of Sodium Salicylate (Procedure 4) (All amounts based on 100 ml. of solution) Peroxide Active Oxygen Ratio of Ferrous Taken, Fe + + + Found, Found, t o Iodometric Mg. M x 105 P.P.M. Active Oxygen CHP, 0.206 2.30 89,200 0.90 AIOHP, 0 . 2 3 2 0.89 30,700 0.71 XIOHP 0 . 6 9 7 2.42 27,800 0.65 3 1 0 H P : 0.697a 2 45 28,100 0.66 a Jt-ith 1.0 ml. of ferrous iron instead of 0.25 ml.
rather than 5 minutes after addition of the iron gives loiv results, owing to the solvolysis of the ferric iron. Procedure 3 appears suitable for the colorimetric determination of cumene hydroperoxide. It appears preferable to Procedure 2 because less ferrous iron is used, so that a given amount of ferric iron in the ferrous solution introduces a smaller blank correction; thus this solution can be prepared and stored with less care. Determination of Various Peroxides. Procedure 3 has been investigated with representative peroxides used in the study of the titration procedure; the data are shown in Table XIV. Procedure 3 gives, with each of the above compounds, active oxygen values that are nearly independent of the amount of peroxide taken. However, the values determined by this ferrous procedure are in general lower than those determined iodometrically ; with methyl oleate hydroperoxide, the average value of the ratio of the active oxygen contents as determined by the two methods is 0.41; with oxidized S.F. fatty acid, it is 0.74; and with oxidized cottonseed oil, it is 0.18. Similarly, in the titration procedure, ratios of 0.20, 0.90, and 0.20 were found (Tables VI1 and VIII). I t appears that induced decomposition of these peroxides is not completely suppressed in either procedure. The low values found by the ferrous procedures could also be accounted for if the acid-catalyzed decomposition of the peroxide competed successfully with the peroxide-ferrous iron reaction. The experiments of the following section were carried out to study this possibility. Reaction at High pH. I n order to minimize the acid-catalyzed decomposition of the peroxide which, in Procedure 3, could take place immediately after addition of the acid solution of ferrous iron, a procedure was developed in which the ferrous iron-peroxide reaction was carried out in alkaline medium, using sodium salicylate to control the pH and to tie up the ferric iron in a soluble complex. Procedure 4. To a solution of the peroxide in about 80 ml. of distilled acetone, in a 100-ml volumetric flask, are added 3.2 ml. of water and 1.0 ml. of a 0.5 JJ solution of sodium salicylate in 95% acetone. The solution is boiled 5 minutes and then 0.25 ml. of a 0.1 111 solution of ferrous perchlorate in 2 M perchloric acid is added. After 5 minutes’ further boiling, 0.5 ml. of 20% perchloric acid is added; the solution is cooled, and 2 ml. of a 14% solution of ammonium thiocyanate in acetone are added. The color is read a t 460 mg after 5 minutes’ development. Compared with Procedure 3, Procedure 4 gives somewhat better results with methyl oleate hydroperoxide, but inferior results with cumene hydroperoxide. With neither peroxide is suppression of the induced decomposition complete. DlSCUSSION
Application of the principles discussed in the beginning of this paper has led to the development of simple procedures giving well reproducible values for the active oxygen contents of various peroxides. Like other ferrous iron procedures which have been developed, however, these procedures give results which, for some peroxides, may differ from the true values by several fold in either direction. The present work has elucidated the source of these errors. Future attempts to develop accurate
V O L U M E 2 3 , NO, 4, A P R I L 1 9 5 1 procedures must, so far as the authors can see, follow the approach used here-namely, to exclude oxygen, and to attempt t o suppress the induced decomposition of the peroxide. Possible means of achieving this suppression would be to use solvents other than acetone, or to add compounds of strong suppressing effect,such as chloride or bromide ions, or maleic or fumaric acids (19).
For general use, where accurate results are desired, iodometric procedures are preferable to ferrous methods at present. However, where some special objection exists to the use of an iodometric procedure, a ferrous method may be of value. Two of the procedures developed above give fairly accurate results with cumene hydroperoxide and soap peroxides, so that investigation of these procedures with various other peroxides would seem desirable. For control work, where the result of the peroxide analysis is correlated with some other property such as rancidity or explosive tendency, and the true peroxide content is of secondary importance, peroxide determinations carried out by the ferrous method in the presence of air are of particular value, owing to the increased sensitivity resulting from the induced air oxidation of ferrous iron. Such methods include the titration procedure in air (above), and various colorimetric procedures dewribed elsewhere ( 1 , 3,20, 26). SUMMARY
The determination of organic hydroperoxides by reduction with an excess of ferrous iron is inherently much less accurate than iodometric methods. For control work ferrous iron methods may be very useful for the determination of traces of organic peroxides. When the reaction is carried out in the presence of an excess of oxygen, the results may be several times greater than the theoretical value, but reproducible results can be obtained. I n the present work the solvent used was acetone, which is a suppressor of the induced decomposition of the peroxide by the ferrous iron-peroxide reaction. Procedures are given in which the excess of ferrous iron is determined by amperometric titration viith dichromate or the ferric iron formed is determined colorimetrically after addition of thiocyanate. I n the absence of oxygen, accurate results are obtained with certain peroxides but not
603 with others. The sources of error of the various procedures are discussed in detail. LITERATURE CITED
(1) Bolland, J. L., Sundralingam, A,, Sutton, D. A , , and Tristram, G. R., Trans. Inst. Rubber Ind., 17, 29 (1941). (2) Dastur, iY.N., and Lea, C. H., Analyst, 66,90 (1941). (3) Golden, M. J., J . Am. Pharm. Assoc., 35, 76 (1946); 37, 234 (1948). (4) Hock, H., and Lang, S., Ber., 77B,257 (1944).
(5) Kharasch, M. R., private communication. (6) Kokatnur, V. R., and Jelling, hl., J . Ana. Chem. SOC.,63, 1432
(194 1). (7) Kolthoff, I. XI., and Harris, K.E., IND.ENG.CHEM.,ANAL. ED., 18, 161 (1946). (8) Kolthoff, I. RI., and Laitinen, H. A., private communication. (9) Kolthoff, I. IT.,and Medalia, A . I., J . Am. Chem. SOC., 71, 3777, 3784, 3789 (1949). (10) Kolthoff, I. M.,and Stenger, V. A., “Volumetric Analysis,” Vol. I, Kew York, Interscience Publishers, 1942. (11) Laitinen, H. d.,and Nelson, J. s.,IND.ENG.CHEM.,ANAL.ED., 18, 422 (1946). (12) Lea, C. H., J . SOC.Chem. Ind., 64, 106 (1945). (13) Ibid., 65,286 (1946). (14) Lea, C. H., Proc. Roy. SOC.,108B, 175 (1931). (15) Lea, C. H., “Rancidity in Edible Fats,” London, 1938. (16) Lips, A,, Chapman, R. A., and McFarlane, W. D., Oil & Soap, 20,240 (1943). (17) May, D. R., Ph.D. thesis, University of Minnesota, 1944. (18) Medalia, A4.I., and Kolthoff, I. M.. J . Poliimer Sci.. 4, 377 (1949). (19) Merz. J. H.. and Waters. W.A.. J . Chem. Soc.. 1949.S15. 120) Robey, R. F., and Wiese, H. K., ISD.ENG.CHEM.,ANAL. ED., 17, 425 (1945). (21) Tanner, E. M.,and Brown, T. F., J . Inst. Petroleum, 32, 341 (1946). (22) TTagner, C. D., Clever, H. L., and Peters, E. D., ANAL. CHEM.,19,980 (1947). (23) Wagner, C. D., Smith, R. H., and Peters, E. D., Ibid., 19,976 (1947). (24) Ibid., p. 982. (25) Woods, J. T., and Mellon, XI. G., ISD.ENQ.CHEM.,ANAL.ED., 13,551 (1941). (26) Young, C. A , , Vogt, R. R., and Nieuwland, J. A., Ibid., 8 , 198 (1930). (27) Yule. J. A. C.. and Tt-ilson. C. P.. Jr.. I n d . Ena. Chem.. 23. 1254 (1931). RECEIVED August 7, 1950. Work carried out under the sponsorship of the Office of Rubber Reserve, Reconstruction Finance Corp., in connection with the government synthetic rubber program.
Determination of Fluorine in Monofluorinated Organic Compounds in Air and Water JASON R I . SALSBURY’, JA-\IES W. COLE, JR., LYLE G. OVERHOLSERI, ALFRED R . ARMSTRONG$, AND JOHN H. YOE University of Virginia, Charlottesville, Vu.
I
T IS generally recognized that i t is more difficult to convert
covalently combined fluorine t o fluoride ion than the corresponding cases with the other halogens. A number of investigators have shon-n that the rupture of the chlorine t o carbon bond may be accomplished fairly readily by refluxing the compound withan excess of an alkalielement in ethyl alcohol (2,10,12,16,19, SS, 36). This approach has been successfully used with a few aryl fluorine compounds and fluorophosphates (21, 38). 3Iost of the reports, however, indicate that more drastic conditions usually requiring involved procedures are necessary for quantitative Present address, dmerican Cyanamid Co., Stamford, Conn. Present address, Carbide a n d Carbon Chemicals Corp., Oak Ridge, Tenn. a Present address, Department of Chemistry, College of William and M a r y , Williamsburg, Va. 1
2
conversion of fluorine in alkyl fluorides t o fluoride ion ( 4 , 1 7 , 2 2 ,Z,?, 24, 27, 28, 35, 42). Kilpatrick ( 1 6 ) has recently discussed the kinetics and mechanism of hydrolysis of some fluoro-organic compounds. When small quantities of tosic fluoro-organic compounds are dispersed in air or dissolved in water, i t is particularly desirable to have rapid and accurate methods for their estimation ( 5 , 6 ) . Several somewhat elaborate methods have been reported. I n gas studies, Cadenbach ( 7 ) , Drake (9),and \T7inter ( 4 0 ) converted combined fluorine t o fluoride ion by burning mixtures of the fluoro-organic conipound with hydrogen a t a jet; the hydrogen fluoride formed was absorbed in aqueous sodium carbonate. Henne (14) detccted organic fluorides in air by the formation of a white cloud when the niisture was passed over silica heated t o incandescence and the effluent gas \vas mised with ammonia.
