The values of An? and Adso were determined from Equations 9 and 10, using the known oxygen contents and the average k: and k: for either aromatic ethers or phenolic compounds. These values were subtracted from the experimental values of refractive index and density. respectively, to obtain the approximate physical properties of the structurally analogous aromatic hydrocarbons. The ring analysis method of Hazelwood (3) for highly aromatic hydrocarbon samples was then applied. The resulting values for total rings, RT, and aromatic rings, RA, are presented in Table 111. The ring analysis values found without taking the oxygen content into consideration are given in parentheses. The known ring contents are given for comparison. DISCUSSION
The ring contents found by the modified method agree well with the known values for both synthetic blends of oxygen compounds and individual constituents. On the other hand, if it is assumed that there is no oxygen pres-
ent, ring contents are found which are very much different from the known values. The small discrepaniSies observed with the modified method are probably as much owing to the inherent limitations of the Hazelwood method (at least for pure compounds and their simple mixtures) as to the small errors inherent in the niodification. In all instances in which it is assumed that there is no oxygen present the values of RT are much too largp, while the Yalues of RA are much too small. With the modified method the values of RT are just slightly too large, while the values of R A are slightly too small. In some instances it may be desired to obtain ring analyses on samples which contain both aromatic ethers and phenolic compounds in an undetermined ratio. Since the effect of oxygen in either form is up to ten times greatei for density than for refractive index, the values of k: are of primary significance. -4 mixture with a small proportion of phenolic compounds would require a k: of about 0.008, whereas a mixture with a large proportion of phenolic compounds would require a k: of
about 0.010. The aromatic ether k:, 0.007, and the phenolic compound k:, 0.011, give an average value of 0.009, This differs by only 0.001 from the values of 0.008 and 0.010 mentioned previously. Therefore, the average k:, 0.009, could be used for an unknown mixture of aromatic, ethers and phenolic compounds. LITERATURE CITED
(1) Am. Petrol. Inst., API Research Project 44, “Selected Values of Proper-
ties of Hydrocarbons and Related Compounds,” Carnegie Institute of Technology, Pittsburgh, Pa. (2) Egloff, G., “Physical Constants of Hydrocarbons,” Reinhold, New York, 19.57.. (3) Hazelwood, R. S . , ANAL. CHEM.26, 1073 (1954). (4)Karr. C.. Jr.. U. S. Bur. Mines. Inform. Ci&. 7802 (19.57). ( 5 ) Sullivan, L. J., ‘Frjes, R. J., McClenahan, IT. S., Willingham, C. B., ANAL.CHEM.29,1333 (1957). . (6) VanXes, K., VanWesten, H. ’A, “-4spects of the Constitution of Mineral Oils,” Elsevier, New York, 1951. ~I
RECEIVEDfor review April 4, 1961. Accepted May 22, 1961. Division of Fuel Chemistry, ACS, St. Louis, Mo., March 1961.
Determination of Hydroperoxides in Hydrocarbon by Conversion to Hydrogen Peroxide and Measurement by Titanium Complexing HARVEY POBINER Analytical Research Division, Esso Research and Engineering Co., linden, N. J.
b A spectrophotometric determination of hydroperoxides in hydrocarbon solution is described. The hydroperoxides are selectively extracted from alkyl and aromatic hydrocarbon solutions by alcoholic caustic. Conditions are established for converting the hydroperoxides to H 2 0 2 . The H202, in turn, is measured b y the titanium-peroxysulfate complex. By reference to calibrations with H 2 0 2 and commercial hydroperoxides, the hydroperoxide content of a hydrocarbon sample is determined a t levels of 5% to 1 p.p.m.
HE presence of hydroperoxides in many hydrocarbon distillates has been associated with sediment formation (fO), odor, and color. It has been suggested (12) that certain low temperature oxidation processes of hydrocarbon reach maximum hydroperoxide levels of 10 p.p.m. Studies of hydrocarbon oxidation are in need of
sensitive measurements of hydroperoxide concentration. *Many of the published ( 7 ) volumetric methods concerned with the reduction of the peroxide linkage often give high blank values which obviate the detection of 1 to 10 p.p.ni. of hydroperoxide. Therefore, a spectrophotometric approach. employing the peroxy-titanium complex. was investigated as a means for achieving this sensitivity. Strohecker, Vaubol, and Tenner (11) and Furmanek and Manikowski (3) used dilute acid to hydrolyze certain peroxides to H202 and determined total H202colorimetrically with a titanium salt. More recently, 4IacSevin and Urone (6) used the titanium complex to remove H202 from aqueous mixtures with R - 0 4 - H . I n this way, they eliminated the current-voltage curve of H202 from polarograms of mixtures of H202 and water-soluble organic hydroperoxides.
It has been reported (9) that tertbutyl hydroperoxide apparently was insensitive to titanium complexing under specific conditions which did complex H202. Other references (2, 11)indicate that organic peroxides do not respond to the titanium complex. This paper reports on the extension of the titanium. method to include organic hydroperoxides. Under controlled conditions different peroxide compounds can be aciddegraded to H202 and measured by the titanium method. Hydroperoxides are acidic and can be removed quantitati~elyby caustic extraction. One chemical reduction method ( 4 ) employs a reflux of dilute aqueous caustic a ith the organic solvent containing hydroperoxide. 11cthanolic KaOH removes hydroperoxides from hydrocarbons 11 ithout the necessit of reflux. The alcoholic caubtic containing hydroperoxide is treated with sulfuric acid a t elevated ternVOL. 33, NO. 10, SEPTEMBER 1961
peratures. This treatment converts R-0-0-H to HZ01 (Equation 1). H202, in turn, is dissociated in the acid solution (Equation 2). Reaction with a titanium salt in H2S04 produces the titanium-peroxysulfate complex (Equation 3). This is a yellow complex and is measured spectrophotometrically a t its absorption maximum, 407 mp, Equations 1, 2, and 3 present the over-all stoichiometry and not a finalized mechanism for these reactions.
ml. with distilled water. Run the second flask is the hydrocarbon blank visible spectrum of each solution us. solution. If solutions are turbid, filter the reagent blank from 700 to 360 mp. on Whatman No. 42 paper. Use 1-cm. cells and a recording spectroRun each of the solutions in the visible photometer, such as the Beckman DK-2. spectrum us. the reagent blank prepared Determine the absorbance a t the maxiin the H202 calibration. For all abmum, 407 mp. Take absorbance a t sorbance curves, measure the absorb700 mp as a reference zero reading. ance a t the maximum (405 to 407 mp), Calculate the grams of peroxide relative to a zero reading a t 700 mp. in each aliquot. Use the HzO2 standCorrect the complexed hydroperoxide ardization value, the 02-2/H202 consolution for any absorbance in the version factor, and the dilution factor in hydrocarbon blank solution. Deterdetermining the grams of 02-2. Plot mine the grams of hydroperoxide Oa-2 the grams of 02+us. absorbance a t 407 by referring to the calibration curve. ROOH His04 + ROSOI OH mp. A tenfold increase in sensitivity is Calculate parts per million of 02+. 67-63'C. realized by the use of 10-cm. absorption (1) cells. DISCUSSION Calibration with Hydroperoxide. H202 + 2H+ 02-2 (2) Prepare a blend of 0.1 gram weighed Extraction of Hydroperoxides from 02-2 Ti+' 2H2S04+ to 0.1 mg., of a standardized hydroHydrocarbon Solution. Hydroper[ T i 0 ~ ( S 0 & ] - ~ 4H+ (3) peroxide in 100 ml. of 1% NaOH in oxides are extracted quantitatively methanol. Pipet aliquots of 5, 10, Absorption max. 407 mp from alkyl and aromatic hydrocarbon 20, and 30 ml. into 100-ml. volumetric solution by alcoholic caustic. The The conversion to hydrogen peroxide flasks. Proceed as in the H202 calidialkyl and diaroyl peroxides are less is quantitative for the hydroperoxides bration, beginning with the addition acidic than hydroperoxides and do examined. The reaction is dependent of the Ti-HzSOp solution. not respond to this treatment. This on heat and acid treating. The comThe symbol O Z - is ~ a convenient way is demonstrated by analyses of the alcoplex is indefinitely stable. It can of correlating calibration data from holic caustic extracts of heating oil measure 1 pap.m. of hydroperoxide peroxy compounds of different molecuoxygen extracted from hydrocarbon. lar weight. The concentration of 02-2 and xylene solutions by the titanium method (Table I). is calculated by multiplying the peroxy Other peroxy compound types, such EXPERIMENTAL compound calibration concentration by as the peracids and the alkyl peroxides, the per cent purity and by a factor of Reagents. COMMERCIAL HYDROcan be acid-degraded to hydrogen 32/mol. wt. peroxy std., where 32 repre. PEROXIDES. tert-Butyl hydroperperoxide in acid solution and then sents the ionic species 0 2 - 2 . oxide, p-menthane hydroperoxide complexed with titanium. Thus, it Analysis of Hydrocarbon Samples (Lucidol Division, Wallace and Tierwas necessary to determine the degree Extract 100 ml. of hydrocarbon samnan, Inc., Buffalo, N. Y.), cumene ple with four 10-ml. volumes of of selectivity introduced by an initial hydroperoxide (Matheson, Coleman 1% NaOH in methanol. If the alcoholic caustic extraction of a sample. and Bell, No. T-7849). hydrocarbon and alcohol phases are OTHERPEROXY COMPOUNDS. Di-tertTable I also indicates that a peracid miscible, as they would be in xylene butyl peroxide and tert-butyl perbenzo(peracetic acid) and a diaroyl peroxide solution, obtain the required phase ate (Lucidol Divisic-n, Wallace and (benzoyl peroxide) were not extracted separation by adding 10 ml. more of Tiernan, Inc.) ; peracetic acid (Food and from hydrocarbon by alcoholic caustic. water with each 10 ml. of alcoholic Machinery Co., Becco, Buffalo, N. Y . ) . It is possible that an alcohol-insoluble caustic. Collect the extracts together Hydrogen peroxide, 30%, Fisher Certisodium salt of the peracid formed and in a 100-ml. volumetric flask and dilute fied, Yo. H-325. Potassium titanium could not be extracted. The diaroyl perto the mark with methanol. oxalate, K2TiO(C204)2. 2H20, Fisher, Pipet 20 ml. of this solution into each oxides are insoluble in alcoholic caustic NO. T-318. and present no interference. However, TITANIUM-SULFURIC ACIDSOLUTION. of two 100-ml. volumetric flasks. To one flask add 10 ml. of Ti-H2S04 and This solution is similar to one that was the perester (tert-butyl perbenzoate) is 25 ml. of 1-1 HzS04, heat a t 57" to used in a previous method (6) for titaquantitatively removed from a blend in 63" C. for 10 minutes, and dilute to the nium. Weigh 10.00 f 0.01 grams of heating oil. Thus, in a mixture of a hymark with distilled water. To the potassium titanium oxalate into a 1droperoxide and a perester, the extracother flask add only the 25 mi. of 1-1 liter beaker. Add 20.0 f 0.1 grams of tion step does not separate these two perheat a t 57" to 63" C. for 10 ammonium sulfate. Add 100 ml. of oxy compound types. Peresters have minutes, and dilute to the mark with concentrated sulfuric acid. Bring to a not been reported as constituents in distilled water. The first flask is the boil in the hood and continue to boil for heating oil. complexed hydroperoxide solution. The 10 minutes to eliminate oxalic acid. Allow to cool. Pour the solution slowly, with constant stirring, into 350 ml. of distilled water. Finally, dilute to 500 ml. with distilled water in a volumetric flask and mix well. Table 1. Recovery of Hydroperoxides from Hydrocarbon Solution METHANOLICCAUSTIC SOLUTION, Analytical NaOH. 1% in CH,OH. Recovery, Gene& Calibration with Hydrogen yo of Theory Blend, Wt. yoin Heating Oil Peroxy Cpd. Peroxide. Prepare a master solution of 1 gram, weighed to 0.1 mg., Hydroperoxide tert-Butyl hydroperoxide 0.22 100 of standardized hydrogen peroxide Cumene hydroperoxide 0.34 95 in 500 ml. of distilled water. Dilute 1.27" 96 6.78" 101 this 50/250 in distilled water and p-Menthane hydroperoxide 0.25 91 pipet the following aliquots into 100Per ester tert-Butyl perbenzoate 0 . 9 8 97 ml. volumetric flasks: 0 (reagent Dialkyl peroxide Di-tert-butyl peroxide 0.36 0 blank), 1, 2, 5, 10, 15, 20, 25, and 35 ml. n- . nfi nPipet in 10 ml. of Ti-H2S04and 25 ml. 0 0.06 Diaroyl peroxide Benzoyl peroxide of 1-1 HzS04. Heat a t 57" to 63" C. 0 1.01 Peracid Peracetic acid (135" to 145' F.) for 10 minutes in a In xylene solution. constant temperature water bath. Cool to room temperature and dilute to 100
1 0 E!
Table It. Stability of Titanium Peroxy Complex Derived from ferf-Butyl Hydroperoxide
-1 J W
Figure 1. Absorbance curves of Tiperoxysulfate derived from tert-butylhydroperoxide Curve
Calibrations with Commercial Hydroperoxides, Peroxides, and HzOz. A series of hydroperoxide calibrants a-as prepared in methanolic caustic, acid-degraded, and complexed with titanium. The resulting titaniurn-peroxy sulfate complex was then measured a t 407 mp. Typical absorption curves of the titanium complex derived from tertbutyl hydroperoxide are ehown in Figure 1. rl Beer's law plot of the 02-2-gram equivalence from all calibration standards (HzOZ, tert-butyl hydroperoxide, cumene hydroperoxide, etc.) us. absorption of the titanium-peroxy complex a t 407 mp is a straight line (Figure 2). This is to be exyected, since Hz02 is the degradation product of the peroxide calibrants. To express the concentrations of the different commercial standards as grams of O2+,it was necessary to assay the standards. The standards of the calibration curve were assayed by a published iodometric method (8). This purity value, together xith the dilution factor and the 02-2/R-O--0-H conversion factor, permitted the calculation of grams of 02-2.
Color Development. Heat soakir?g the complex is the critical step in the color development. Literature referenceb (8, 2.2) indicate that organic peroxides do not respond to conditions which complex HzOs. tert-Butyl hydroxide was reported (6) to respond a t very low sensitivity, 1/500th the absorptivity of an equivalent amount of H202. This low sensitivity may be due to incompleteness of the conversion of the hydroperoxide to Hz02a t room temperature (Equations 1 and 2). For example, when the concentration of H2S04was varied from 4.1 to 11.7N, the absorption coefficient of the complex a t 407 m p varied from 0.17 to 2.3, a value about one fourth that of the stable Coefficient for the complex. l'hi.; low sensitivity and instability
Spectral Measurement Run immediately After 19 hours After 7 days After 24 days
With heat soaking, 135' F. 8.46 8.44 8.35 8.34
Without heat soaking
G. 0 2 - 2
Figure 2. standards
were eliminated by a heat soaking step. Heat soaking a t 57-63' for 10 minutes is suitable for forming a stable complex. These stability data, as in Table 11, were taken as indicative of completion of the reactions of Equations 1, 2, and 3. The correlations of different calibration standards in producing equivalent amounts of H202were 'additional proof of the completion of reaction under these conditions. There is complete stability with time (0 to 24 days) when heat soaking is used, as opposed to weaker absorbances and instability (increasing absorbances with time) when heat soaking is not used. The limiting range of 57' to 63' C. (135' to 145' F.) was selected as that at which the methanol of extracted samples would not appreciably vaporize. Repeatability and Accuracy. At the 10-p.p.m. level of hydroperoxide 0 2 - 2 in heating oil, the sigma is =k0.8%. At the 5 weight % hydroperoxide O Z - level, ~ the sigma is j=O.l%. These figures are based on five replicate determinations. Accuracy was indicated by the 95 to 100% recovery of hydroperoxide from synthetic blends in heating oil and in xylene by the method developed. Interferences. Interferences in the method are a function of the formation of H20zby other substances undergoing acid degradation under the conditions of the method, and the reactivity of the hydroperoxide with other compound types in a mixture. The formation of H202 by other organic peroxy compounds is described in the discussion of the estraction technique. A perester is an example of an interfering peroxy compound type. Representative hydrocarbon, oxygen, sulfur, and nitrogen compound types do not respond to the titanium complex. However, there are some deviations from full quantitative recovery of hydroperoxide in blends with certain compound types, due to chemical interaction before complexing. For example, a hydroperoxide reacts with an aliphatic sulfide (1) to form a dialkyl sulfoside, R2S0, and will complex with
Calibrations with peroxy HzOz fert-buty100H cumeneOOH p-menthaneOOH lert-butylperbenzoate
amines and pyrroles. Thus, the level of hydroperosides in a mixture is dependent on the degree of interaction with other compound types before analysis. Table 111 shows the degrees of recovery of hydroperoxide from blends with osygen, nitrogen, and sulfur types.
Table 111. Dependence of Hydroperoxide Level on Interaction with Other Compound Types
Impurity Acetic acid Oleic acid 1-Octanol Formaldehyde Butyraldehyde Tetrahydrofuran Ethyl acetate Methyl ethyl ketone %Hexanone
Butyl OOH/ Moles Impurity'
Analytical Recovery of Hydroperoxide Complexby Ti
1-1.83 1-15.2 1-0.53 1-1.05 1-1.06 1-2.12 1-3.06 1-6.13 1-0.97 1-1.63 1-3.26 1-1.26 1-2.52 1-1.30 1-2.60
m3 97.1 47.6 99.2 73.7 93.9 75.7 98.1 87.8 93.0 86.8 53.0 95.1 92.1 99.7 93.9
1-1.26 1-2.52 1-1.03 1-2.07 1-0.74 1-2.46 1-1.48 1-1.08 1-1.25
93.4 76.8 88.8 75.9 87.0 4.7 6.7 99.6 2.2
n-Heptyl sulfide Thiacyclopentane Thiacyclohexane n-Butyl disulfide 1-Octanethiol Phenyl p-tolyl 1-0.92 sulfone 99.6 Thio hene 1-2.70 99.5 97.4 PyriXneb 1-1.86 1-1.53 Piperidineb 90.0 1-2.16 Pyrroleb 30.0 a General level of hydroperoxide and im urity was 2 to 4% in heating oil. ?Blends were made with cumene hydroperoxide.
VOL. 33, NO. 10, SEPTEMBER 1961
“Textbook of Quantitative Analysis,” 742. Macmillan. New York. 1948. r6j Mac’Nevin, W . ’ M., Urone, P. K., ANAL.CHEM.25, 1760 (1953). (7) Martin, A. J., in “Organic Analysis,” J. Mitchell, Mitchell, E. J. Proskauer, I. M. A. Weissberger, eds., Chap. 1, Kolthoff, A.’ pp. 2-64, Interscience, New York, 1960. (8) Ricciuti, C., Coleman, J. E., Willits, C. 0.. O., ANAL.CHEM.27.405 27,405 (1955). (9) Satterfield, C. N., ‘Bonnel, Bonnel, A. H., D.
(1) Denisov, E. T., Emanuel, N. M., Uspekhi Khim. 4, 365 (1958). (2) Egerton, A. C., Smith, F. L., Ubbelohde, A. R., Trans. Rov. SOC. (London) ~ 2 3 4 ,433 ‘ (i935). (3) Furmanek, C., Manikowski, K., Roczniki Pahstwowego Zakladu Hig. 4,447 (1953) ; C. A . 48,8559 (1954). (4) Kolthoff, I. M., Meehan, E. J., Bruckenstein, S., Minato, H., Microchem. J. 4, 33 (1960). ( 5 ) Kolthoff, I. M., Sandell, E. B.,
Ibzd., 27, 1174(1955). (10) Sauer, R. W., Weed, A. F., Headington, C. E., Preprint, Division of Petro-
leum Chemistry, ACS, 3, No. 3, 95 (1958). (11) Strohecker. R.. Vaubol. R. V.. ‘ Tenner, A., Fette ‘und Seifin 44, 246
(1937). C. A . 32,816 (1938). (12) Waish, A. D., Trans. Faraday SOC. 42,271 (1946).
RECEIVEDfor review April 26, 1961. Accepted June 29, 1961. Conference on Oxidation of Hydrocarbons in the Liquid Phase a t Low Temperature, U. S. Bureau of Mines, Bartlesville, Okla., May 23, 1961.
A New Falling Velocity Method of Density Determination for Small Solid Samples A. S. ROY Bell Telephone Laboratories, Inc., Murray Hill, N. J.
b A new method is described for determining the density of a solid sample of arbitrary shape and small size b y measuring its fall time in two viscous liquids of different densities, measuring the fall time of a sample of known density, and calculating the unknown density from two equations. This method is not restricted to measuring densities of solids with densities less than liquids available for flotation. The method has been investigated experimentally and found to measure densities of materials from approximately 2.5 to 10.5 with good agreement with reported densities.
HE SIQNIFICANCE of density measurements in solid state studies has been demonstrated recently by Smakula (9) and Horn (6). Frequently only small samples are available for investigation (3, 10) and density determination becomes difficult, as when dealing with single crystals, whiskers, isotopes, sintering, or hot pressing. Classical methods of measuring density have some limitations with respect to small samples. Weighing the sample and determining its volume, either directly, by measuring its dimensions, or indirectly, by substitution in fluid, auffers from loss of accuracy as the sample becomes smaller because of constant errors that do not decrease with sample size. The flotation method in which the sample is floated in a liquid of matching density, though suitable for small samples, is not applicable to samples having a density greater than 4 to 5 grams per ml., because no suitable liquids are available as yet. The falling velocity method, in which the gravitational velocity of a body in a viscous liquid is measured, has the advantage that the velocity of a small body
can be determined as accurately as that of a large one. However, the body must a spherehave a specific shape-.g., and it is necessary to measure its dimensions, which again render the method inaccurate for small samples. These difficulties are removed by the method described below. PRINCIPLE OF METHOD
Barr (1) has shown that the limitations of Stokes’ law for determining the viscosity of a liquid can be practically eliminated when two falling velocity experiments are conducted in two different viscous liquids on the same spherical body under the same geometrical conditions of the fall tubes. Stokes’ law states: v = 2/9
a sphere in Equation 1)for an irregularly shaped particle does not need to be known since it cancels. This cancellation stems from the fact that in laminar flow any arbitrarily-shaped body has a certain shape factor which is a function of its geometry and orientation alone (1, 6, 7 ) . However, in order for Equation 2 to hold, the arbitrary body has to fall with the same orientation in the two experiments. Except for bodies with 3 axes of symmetry, bodies with arbitrary shapes tend to fall with one preferred stable orientation (4, 6) ; therefore, the condition for extending Equation 2 to arbitrary shapes is generally realizable. Thus, a generalized Stokes’ law equation can be formulated for a body of arbitrary (finite) shape falling along its stable orientation:
where u, p, and u are, respectively, the velocity, density, and the radius of the falling sphere; and u are the viscosity and the density of the liquid; and g is the acceleration due to gravity (all in cgs units). In another liquid of viscosity p’ and density u’ a similar relation holds. Hence ( 1 )
in which t is the fall time along a certain path in one liquid and t’ that in the other liquid. The wall-effect and endeffect factors required to correct Equation 1 for applying Stokes’ law to finite media thus cancel out, since they are a function of only the geometrical features of the system (the sphere and the tube) in the case of extreme laminar flow ( I ) . It is significant to note that Equation 2 is independent of the shape of the particle. The shape factor (2/9 u2 for
in which K is a constant characteristic of the geometry and orientation of the body only, and one can obtain Equation 2 for a body of arbitrary shape from Equation 3 applied twice in two different liquids. If two separate pairs of such experiments are performed with two bodies, i and j, two equations of the type of Equation 2 are obtained, one with p,, t,, and t l and the other with p i , t,, and ti in which the subscript denotes the body in question. Dividing one of these equations by the other eliminates the viscosity ratio and results in