V O L U M E 25, NO. 6, J U N E 1 9 5 3
923
LT-7 117 L 7
ro
RELAY
mj.
barostat which would make it possible t o provide a restorative actio’n proportional in some manner to the deviation from control point. Such a scheme, however, arould make necessary additional electronic equipment as well as auxiliary pressure and vacuum sources. It was not found necessary to employ any of the above-mentioned modifications in the controller which was used to maintain constant pressure for the measurement of vapor-liquid equilibrium. A sensitivity of & O . l mm. of mercury without excessive cycling was easily obtained if relative pressures and vacuums of from 5 to 10 mm. of mercury were used.
SWITCH
CONCLUSIONS
Figure 4.
Wiring Diagram of Water Bath Controller
CL. 20 microfarads, 150 v o l t s
Ra.
3400ohms
RI.
Ra.
22,OOO ohms
RI. 60,OOOohmn
600,OOOohms
R,. 10,OOOohms
retaining a high sensitivity within a minimum of cycling would consist of additional electrodes placed in cascade fashion in the
The controller described proved to be reliable and over a period of several months of almost continuous operation, it was found to require but slight adjustment. Although the use of glass as a material of construction limited this controller to pressures in the vicinity of 1 atmosphere, a far wider control of pressure could have been obtained by suitable substitution of metal for glass. RECEIVEDfor review June 2,1952. Accepted March 28, 1953.
Determination of Hydroperoxides in Petroleum Products D. C. WALKER AND H. S. CONWAY Research Department, Standard Oil Co. (Indiana), Whiting, Znd. In a new method, hydroperoxides in petroleum products are determined by extraction and reduction with sodium arsenite. For gasolines and distillate fuel oils the method is more sensitive than the iodometric method, and avoids the dependence o n sample size characteristic of the ferrous method. Known concentrations of hydroperoxides in gasolines, fuel oils, and white mineral oils are determined with an accuracy of 0.03 hydroperoxide number (milliequivalents of active oxygen per liter of sample) for numbers less than 1, and within 2% for higher hydroperoxidenumbers.
T
HE determination of low concentrations of peroxides in
petroleum products has been of concern to the petroleum industry for many years. Formation of gum and sediment in refined distillates, development of disagreeable odors of white mineral oil, and bearing corrosion in engines are among the many effects attributed either directly or indirectly to peroxides. These may form whenever hydrocarbons are exposed to air, and the primary products of the reaction are alkyl hydroperosides (ROOH) rather than dialkgl peroxides (ROOR) ( 3 , 4,9 ) . Published methods for the determination of hydroperqxides utilize either ferrous or iodide ions in acid solution as reducing agents. I n the ferrous method, the amount of hydroperoxide is determined by measuring either the ferric ion produced or the excess ferrous ion. I n t,he iodide methods the liberated iodine is titrated with thiosulfate. The variations of both methods have been adequately summarized by Kolthoff and Medalia ( 6 ) , who concluded that the ferrous method was “inherently much less accurate than iodometric methods.” Several iodometric procedures ( 2 , 6, 7 , 11) are accurate for concentrated hydroperoxides, but no data have been reported on synthetic solutions of hydroperoxide in hydrocarbons more dilute than 0.1 A’. In applying the iodometric methods to petroleum fractions, part of the liberated iodine may be lost by addition to olefinic components or by volatilization ( 8 , 1 2 ) . Arsenious oxide is used as the reducing agent in a method proposed by Siggia ( 1 0 )for the assay of benzoyl peroxide. Because Siggia’s procedure requires complete solution of the sample in aqueous alrohol, the amount of petroleum product that can be
taken for analysis is limited, and low concentrations of hydroperoxide cannot be determined. In a new method that has been developed for the determination of hydroperoxide in petroleum products, aqueous sodium arsenite is used as the reducing agent in a two-phase system, and organic materials are removed by separation and extraction before determination of excess arsenite. Because the sample need not be dissolved in the reagent, there is no limit on the amount of sample that may be taken, and very low hydroperoxide concentrations can be accurately determined. METHOD
The extraction apparatus, a modification of a commercially available tetraethyllead extractor ( I ) , is diagramed in Figure 1. Standard Solutions. Arsenious oxide, 0.1 N . Dissolve 4.9450 grams of pure, dry arsenious oxide in 50 ml. of 1 N sodium hydroxide. Make neutral or very slightly acid to litmus paper with 1‘V sulfuric acid and dilute t o 1liter. Iodine, 0.05 N . Dissolve 12.7 grams of C.P. iodine and 40 grams of C.P. potassium iodide in 25 ml. of water, and dilute to 2 liters. Standardize against the arsenious oxide solution. Procedure. Into the separatory funnel of the extraction apparatus introduce 75 ml. of distilled water, 2.0 ml. of 1 N sodium hydroxide, and 2 drops of phenolphthalein indicator solution. I n accordance with the anticipated hydroperoxide content, add the amounts of sample and of 0.1 N arsenious oxide given in Table I. If the amount of sample used is less than 50 ml., add enough petroleum ether to bring the volume of the top phase to between 50 and 100 ml. Drain the mixture into the lower part of the apparatus and rinse the funnel with 25 ml. of 95% ethyl alcohol. Bubble nitrogen through the solution a t a moderate rate for 5
924
ANALYTICAL CHEMISTRY
minutcs, then turn on the heater. When refluxing begins, adjust the flow of nitrogen and heat for maximum reflux. If the pink phenolphthalein color disappears during reflux, restore the alkalinitr 15 ith 1 -Vsodium hydroxide and add 2 ml. in excess. Reflux for 45 minutes, then acidify the mixture dropwise with 10 S sulfuric acid. Turn off the heater, and cool the extractor to approximately room temperature. Shut off the nitrogen and a l l o ~the phases t o separate. Withdraw the lower phase into a 250-ml. scparatory funnel, extract three times v i t h 25-ml. portions of chloroform, and discard the chloroform washings. Transfer the solution to a 250-mi. beaker and chill in ice. Add 1 gram of sodium bicarbonate and titrate the solution w-ith the standard iodine solution. Determine the end point electrometricallr or v i t h starch indicator.
Table I.
Sample Size and Reagent Volume
Anticipated Hydroperoxide h-uinber" 1to 5
5 to 10 t o 50 t o 100 t o a
MI.
111. of 0.1 N A s s 0 3
100.0 50.0 25.0 10.0
5.00 7.00 7.00 15,oo 15.00 15.00
of Sample
0 to 1 10 50 100 1000
5.0 0.50
Milliequiralents of active oxygrn per liter.
Table 11. Analysis of Hydroperoxide Concentrates Hydroperoxide Tetralin 1,3-Dirnethylcyclopentane Cu niene tert-Butyl Mixture of above hydroperoxides
Active Oxygen Found, 1.75 6.37 8 . 07 11.21 6.70
Z
I.D. 1.8 CM.
Table 111. Hydroperoxide Numbers of Synthetic Solutions
Calculated 0.05
0.10 0.50 0.70 1.00 3.00 5.00 15.0
HEATiNG ELEMEN?
Figure 1.
Extraction Apparatus
-
Run a hlank determination with each saniplr or v i i e s of tmples, using the same amount of reagent and solventq ('alculate the hydroperoxide number by the equation: Hydroperoxide number =
1000 S ( b -
S)
U
where the hydroperoxide numher is expressed as Inilliequivalents of active oxygen per liter, b = milliliters of iodine solution used for blank, s = milliliters of iodine solution used for sample, N = normality of iodine solution, and u = milliliters of sample used. If b is less than 3 times as large as ( b - s), repeat t'he determination in accordance Tyith the proportions of sample and reagent given in Table I. For hydroperoxide concentrates it is more convenient to weigh the sample and to calculate per cent active osygen as follows:
% active oxygen
=
8
.
10 20
Found Fuel Oil Iodide Srsenite (6)
Gasoline Iodide Arsenite (5) 0.06 0.00 0.12 0.00 0.47 0.00 0.71 0.02 0.97 0.06 2.96 1.11 5 03 1.73 15.1 5.46
0 0 0 0 1
3 4 I5
06 12 49 69 02 02 99 0
1
hIinera1 OilIodide Arsenite (a)
0.00 0.00 0.05
0.04 0.10 0.50 0.70 0.99 3.00 4.97 15.0
0.08 0.10 1.93 3.41 12.9
0 0 0 0 0 3 5 14
91 01 03 $1
are the averages of triplicate determinations differing from enrh other by no more than 0.03 hydroperoxide number or 2% of the. amount present, whichever was larger. The synthetic solutions were prepared from the four hl-droperoxides listed in Table I1 x-hich had been chosen to simulate the hydroperoxides formed by oxidation of petroleum products. Approximately equal amounts of the hydroperoxides were mistd to form a concentrate having a calculated active oxygen content of 6.70%. This value was checked within 0.04% both hy the arsenite method and by Kokatnur and Jelling's ( 5 ) iodometric procedure. Known amounts of the mixture xere added to an inhibited cracked gasoline, a distillate fuel oil from high-sulfur crude oil, and a white mineral oil to give calculated hydropcrosicle numbers from 0.05 to 15.0. Each sample was analyzed by tht. arsenite method and by the iodomctric procedure. Table I11 indicates that the arsenite method determined thcx hydroperoxides in all cases with an accuracy of 0.03 for hpdroperoxide numbers less than 1, and within 2% for numbers greater
S)
t v
- BUTYL
-
xhere w = grams of sample used. r\
CUMENE
0 1,3 - DIMETHYLCYCLOPENTANE 0
EVALUATION OF METHOD
The accuracy of the arsenite method x a s tested by the analysis synthetic solutions of hydroperoxides in gasoline, distillate fuel oil, and white mineral oil. For comparison, analyses were also c:irried out by the iodometric procedure of Kokatnur and Jelling ( . 5 ) , which is claimed to be applicable to the determination of "peroxygen in mineral oils," and was used as a reference procedure by Kolthoff and Medalia (6) in their study of the ferroushydroperoxide reaction. The present method was further evaluatcd by application to air-oxidized gasolines and mineral oils, a r i d hy comparison of the results with those obtained by an iodometrir prorc4urr and a ferrous procedure. h l l data reported
0.; 10 .iO 67
I-
-
,TETRALIN
0
5
0
I
I
IO
15
L
ARSENITE : HYDROPEROXIDE RATIO
Figure 2.
Effect of Arsenite-Hydroperoxide Ratio
on Determined Active-Oxygen Content
'
925
V O L U M E 2 5 , N O . 6, J U N E 1 9 5 3
2.0
In the routine use of the arsenite method, occasional samples of gasoline have been encountered which gave negative hydroperoxide numbers indicating the presence of a reducing agent in the gasoline. These gasolines also showed unstable end points during titration of the excess arsenous oxide with iodine. Plausible results could be obtained, however, by titrating rapidly to the f i r p t indication of the end point, because the reaction of the iodine with the reducing agent appears to be considerably alouer than with the argenite.
w t x
1.0
2.5
0
I+
IO 20 30 40 50 ARSENITE : HYDROPEROXIDE RATlO Figure 3. Effect of 4rse lite-Mydrnperoxide Ratio on Hydroperoxide Niiniber of Gasolines
2.0
0
U W
m
5
3
z
1.5
W
thiin I , For gasoline :inti f u c ~ loil, thc iodonietric method failed to r l ~ t c ~low t hydroperoxide nunil)~rs,and gave extremely 1 0 ~ values at higher concentrations. For ivhite oil. however, both niclthods agreed within an average of 0.05 with the calculated valuep. Possibly types of compounds that can consume iodine Lvere present in the gasoline and fuel oil, but had been removed froin the. white mineral oil by the uwal drastic treatment with fuming sulfuric acid. The applicability of the aiwnite method to typical samples of a g d , catalytically cracked gasolines is shown in Table II-. (:oniparative results are given for Wagner, Smith, and Peters' iotlometric~procedure (11) designed for use with hydrocarbons in thc gasoline boiling range, and for Tule and Wilson's ferrous procwlure (f3),Xvhich is estensively used in the petroleum industry. Rwults by the arsenite method are almost always higher than thow ohtained by either of t h r other methods. I n view of the i,esults in Table 111, thip finding as not entirely unexpected. In :I Yiniilar comparison of the aryenite and ferrous methods :il)i)lied to white mineral oils that had stood 2 months in sunlight and air, the arsenite method determined a t least three times as much hydroperoxide a3 did the ferrous method. DISCUSSION
Khilcl :Lt k a s t three equivalents of arsenous oxide must be for each equivalent of hydroperoxide, larger amounts of ursrnics do no harm. Figure 2 s h o w that for hydroperoxide conc.enti,ates the active oxygen contcsnt does not deviate from the ilvc>i'age1))- more than 2% w e n in the pre3ence of a tenfold excess of arsenite. For air-oxidized gasolines, Figure 3 s h o w that the avri'age hydroperoxide number is similarly constant within 0.03 even if a twentyfold excws p l reagent is t,aken. Evidently natur:il hydroperosides react it:: completely with arsenite as do syiith,tir. 1 wkvn
'I'ahle I \ - .
Hydroperoxide Numbers of Aged Catalytically Cracked Gasolines Arsenite
Method Iodide ( 1 1 )
l'errous ( 1 3 )
Uninhibited 0.25Q 0.94 1.04 2.39 5,44
0.20 0.22 0.22 0.37 0.38
0.02" 0 54 0 31 I . 52
0.15a
0.64 0.60 1.61 2.56
0.95 Inhihited (0.002% U.O.P. 5 ) 0.25 0.08 0.00 0.24 0.10 0.08 0.94 0.36 0 31 0 17
e ia3 W
1.0
a a
>.
I 0.5
n
"0
20 30 40 OXIDATION TIME, HOURS
IO
50
Figure 4. Relationship of Hj-droperoxideNuniber to Degree of Oxidation of a Distillate Fuel Oil
S o difficulty has been encountered in the applic:atioii of the irsrnite method to distillate fuels and to waxes. Hydroperoxide numbers obtainrd on successive samples of distillate fuel oil, air-oxidized at room temperat,ure, are plotted in Figure 4. The method can be used as a measure of the degree of oxidation of fuel oil. In \vas analysis the sample is weighed rather than measui,ed by volume, and the hydroperoxide number is then calculated as milliequivalents of active oxygen per kilogram of wax. ACKNOWLEDGMENT
The authois are indebted to the laboratorieq of the Cniveiwl Oil Products Co. for the samples of aged gaqoline. LITERATURE CITED
(1i dni. So?. Testing Naterials, Philadelphia, "ASTJI Standards,"
Part 5 , Designation D 526-48T, 1949. (2) Dastur, S . S . ,and Lea, C . H., A n a l u d , 66, 90 (1941). (3) Ellis, C., "Chemistry of Petroleum Derivatives," Vol. 11, Chap. 40, Sew York, Reinhold Publishing Corp.. 1945. (4) Farmer, E. H., Trans. Faraday Soc., 38,340 (1942). ( 5 ) Kokatnur, V. R., and Jelling, AI., J . Am. Chem. Soc., 63, 1432 (1941). (6) Kolthoff, I. XI., and Nedalia, A. I.,ANAL.C H E X . ,23, 595 (1951). ( 7 ) Lea, C. H., J . SOC.Chem. Ind., 65,286 (1946). (8) Panyutin, P. S., and Gindin, L. G., BUZZ. acnd. sci. C.R.S.S., Clnsse s e i . ninth., Sir. chim., 1938, 841. (9) Sachanen, -1. S . , "Chemical Constituents of Petroleum," Chap. 7 , ?;ew York, Reinhold Publishing Corp., 1945. (10) Siggia, S.,-AWL, CHEM.,19, 872 (1947). (11) Wagner, C . D., Smith, R. H., and Peters, E. D., Ihid.,19, 976 (1947). (12) Widniaier, O., and XIauss, F., Rev. inst. franc. pbtrole, 3, 18.3 (1948). (13) Yule, .J. .\, C., and Wilson, C. P., Irid. E n y . Ciieni., 23, 1254 (1931). R E C E I V E fDo r review December 1 1 , 1952. dccepted March 9 . 1953. Presented before the Division of Petrokuni Chenii-try a t f l i t . 1 2 3 r d l l ? r . r i r n of the AMERICAS C H E M I C ASOCIETY. L Los Ange1e.s. Calif.
