Another set of pore-diameter data n a s published by Wheeler ( 6 ) , and is given in Table T.’. EFFECT OF ADSORBENT PORE SIZE ON THE SEPARATION OF PETROLEUM RESIDUES
Pores larger than 22 A. in diameter are required for the efficient adsorption of the resinous and asphaltic components of oil residues. Other tests, made in this investigation, indicated that Attapulgus clay or alumina (Alcoa F-20) used, above grade 912 silica gel in a multiple adsorbent column, m-as effective in adsorbing the colored portions of typical residues. The procedure finally adopted for use employs the two types of silica gel. Silica gel (grade 912) or another grade n-ith 2 2 4 . pores is the best adsorbent for the retention of the simpler aromatic molecules. The use of tlTo types of silica gel in a multiple adsorlient column will eliminate the catalytic surface reactions which might occur if alumina or clay is used as a second adsorbent. Comparative tests on typical residues TI ith cyclohexane, benzene, and methanol as eluents, indicated that a greater recovery of material was obtained from the multiple silica gel column than from the clay, or alumina and silica gel (grade 912) combinations.
If it should appear, that undesirable catalytic surface reactions occur on the surface of the silica gel, it might be interesting to evaluate a multiple colunin containing a bed of large-pore alumina above a section of small-pore alumina (about 20-rl. pore diameter), -4t present such small-pore alumina is not aw,ilable. The adsorption results of O’Donnell ( 2 ) may be explained, on the basis that the heavier fractions contain increasing amounts of resins and asphaltenes. Because small-pore silica gel has a low capacity for these components, larger adsorbent-to-sample ratios were required. The addition of alumina, which has larger pores, to the top of the silica gel column greatly improved the retention of asphaltenes and resins. CONCLUSION
The proposed method for the preparation of saturate concentrates of oil residues has proved adequate for present purposes, because all of the prepared saturate concentrates d i i c h lvere analyzed contain less than 5% aromatics. I n other work, where even l o w r aromatic content is required, other solvents, adsorbents, eluents, or choice of cut point may yield better results. The solution of a total oil residue in a cyclic hydrocarbon solvent, followed by elu-
tion chromatography on a multiple column containing a large-pore adsorbent above a small-pore adsorbent, appears to be a generally useful method for the analysis of heavy petroleum materials. The fact that adsorption is controlled to a considerable degree by adsorbent pore size facilitated the development of a direct method for the separation of the saturated hydrocarbons from the oil yesidues. This method, unlike others ivhich have been proposed, does not require a preliminary removal of asphaltenes or resins. LITERATURE CITED
(1) Kleinschmidt, L. R., J . Research -V:atl. Bur. Standards 54, 163-6 (1955). ( 2 ) O’Donnell, G., *1ln-.4~. CHEII. 23,
894-8 (1951).
(3) Ries, H. E., Jr., “Advances in Catalysis,” Vol. ITr, pp. 88-119. .Icademic
Press, Xew York, 1952.
(4) Schxeyer, H., Chelton, H., Brenner,
H. H., Proc. Assoc. Asphalt Pating
Technol. 24, 3-30 (1955). ( 5 ) Shell Oil Co., Houston Refinery Re-
search, unpublished work.
(6) Kheeler, A,, “Catalysis,” Vol. 11, p. 110, Reinhold, Yew York, 1955.
RECEIVEDfor revieiv >\lay 22, 1958. -1ccepted July 29, 1958. Presented in part before Division of Analytical Chemistry, 131th Meeting, -ICs,,Chicago, Ill., September 1958. Publication No. 170, Shell Development Co., Exploration and Production Research Division, Houston, Tex
Determination of Dioctyldiphenylarnine in Hydraulic Fluids S. W. NICKSIC and S. H. JUDD California Research Corp., Richmond, Calif.
b The estimation of p,p’-dioctyldiphenylamine is required for manufacturing control and oxidation studies on hightemperature, silicon-based hydraulic fluids A method has been developed based on the reaction of the amine with furfural in the presence of strong sulfuric acid. Degradation products of the amine and the fluid d o not interfere. The reaction with furfural i s apparently a general one for primary and secondary aryl amines, as shown b y the reaction with 15 selected compounds. Similar procedures may b e developed for other aryl amines individually and in the presence of nonaryl amines.
G
performance is a n important prerequisite for certain aircraft hydraulic fluids. A OOD, HIGH-TEMPERATURE
2002
ANALYTICAL CHEMISTRY
composition of growing commercial interest consists of a specific disiloxane derivative with a silicon thickener and p,p’-dioctyldiphenylamine (DODA) as an oxidation inhibitor. The determination of this inhibitor is desirable for manufacturing control and for correlation with the degree of degradation of used hydraulic fluid. Total nitrogen can be determined only with difficulty, because at low levels the blank values approach those for the samples. Further, the nitrogen test does not distinguish the inhibitor from its degradation products. Laboratory oxidation studies have been used, but they are not rapid or sufficiently specific. This paper describes a colorimetric test for determining DODA either in hydrocarbons or in material containing derivatives of silicon. Degradation products of
the fluid and of the inhibitor do not interfere. The reaction of furfural with amines !vas considered in early exploratory studies. With most primary amines the reaction forms Schiff’s bases, RC=XR’;but when the amine is aromatic, more complex reactions can take place ( 3 ) . Aromatic amines, both primary and secondary, give highly colored products in the presence of strong sulfuric acid. Table I illustrates the colors produced with a number of amines. These colors are obtained by treating 35 ml. of a 2-propanol solution of amine successively with 10 ml. of 18S sulfuric acid and 5 ml. of 10% furfural in 2-propanol. Colors of various hues are obtained whenever a primary or secondary amino group is attached directly to an aromatic ring. The only tertiary amine
2 200 2 00.i I
800
A C
CODA 5 - A N C A R D hEW OIL
- SAMPLES ---- 9 - A N K S
/
I
I I0
I5
20
MILLILITERS 10N HrS04 IN 50-ML SAMPLE
Figure 1. Acid concentration v5. absorbance after 60 minutes
tested, dimethylaniline, gave no color. The color reaction with DODA is not intense, but the test is sufficiently sensitive for the amounts normally found in hydraulic fluid.
MC MG
MG
METHOD
.AhK
Apparatus. Becknian Alodel B spectrophotometer or photometer n i t h a n a r r o v band-pass filter in t h e 380to 400-mp range. Reagents. Celite, S o . 503 or S o . 535, Johns-Jlanville Corp. p,p'-Dioctyldiphenylamine, B. F. Goodrich Chemical Co.. A403191; assay 99%+ by nitrogen analysis. Prepare a solution of 1.00 mg. per ml. in 2-propanol. Furfural, Eastman K h i t e Label. Prepare a solution of 10% v./v. in 2propanol using a freshly distilled heart cut. This solution is stable for 60 days a t 40" F. Standard Curve. R u n standards n i t h each set of samples exactly as t h e samples are run. Prepare t h e standards using 2, 5, and 10 ml. of DODA solution diluted to 40 ml. with 2-propanol in SO-ml., graduated, glass-stoppered cylinders. Treat as described below. Procedure. K e i g h a 0.2-gram saniple into a 50-nil. glass-stoppered graduate, a d d 40 ml. of 2-propanol, 10 ml. of 1 6 s sulfuric acid, and mix. Use caution, as heat is produced. Add 0.5 gram of Celite and filter through a glass wool plug. Discard the first few milliliters and collect enough clear filtrate to yield two 10-ml. aliquots. The mixture may be centrifuged for 15 minutes a t 2000 r.p.m. if preferred. Aliquot two 10-ml. portions of the filtrate, adding 1.00 ml. of 2-propanol to one for a blank and 1.00 ml. of furfural solution to the other. Read each sample (including standards) against its blank in 30 + 3 minutes at 385 mp in a 1.00-cm. cell. Prepare a standard curve, and calculate the per cent DODB. DISCUSSION A N D STUDY
OF
VARIABLES
Figure 1 shows the variation of absorbance 1%-ithacid concentration at 60
lem, discussed below, become the limiting factors. Figure 2 shows the effect of reaction time on the absorbance. The exact time selected is not critical if the same time is used for both samples and standards. The 30-minute time was chosen because it is a t the flattest part of the curve, and it allows time for preparation of samples and standards for reading a t the proper interval.. The color intensity increases rrith temperature. Room temperature \vas used because heating requires close control to maintain reproducibility, and the gain in sensitivity does not justify added manipulations. The chemical reaction involved does not go to completion, so that standards should be run with each ,set of samples. Care must be taken to ensure that the details of the procedure, such as rate of adding reagents and methods of mixing, are the same for both sample and standards. Figure 3 gives the spectrum of the colored product in the range from 350 to 540 mp. Because the absorbance is measured in a region \\here it changes rapidly n ith wave length, careful attention is required for good reproducibility. The best compromise for IOK blank and high sensitivity is 385 nip for the Beckman AIodel B spectrophotometer used in this laboratory. A photometer with a narrow band-pass filter peaking between 390 and 400 mp should give satisfactory results. A typical calibration curve is slightly nonlinear a t concentrations above 0.2 nig. per ml. A decrease in sensitivity is observed ivvlien furfural reagent that has'been stored for 1 month is used in place of freshly distilled reagent. This decrease may be caused by a decrease in furfural concentration. illthough the amount of furfural added in the procedure is many times the stoichiometric amount, more color is obtained in a given time if more furfural is used. IYith a 5-mg. DODA standard, roughly
040 0 200
350
300
430
470
610
550
WAVE LENGTH I N MILLIMICRONS
Figure 3. Spectra of pure DODA after color development
minutes. The midpoint of the curve was selected as the optimum concentration, with enough absorbance difference between sample and blank for the required sensitivity. Greater amounts of acid give higher sensitivity, but sample solubility and a haze prob-
Table
I.
Reaction of 2-Propanol Solutions of Amines with Furfural in Presence of Strong Sulfuric Acid
Amine S, S '-Diphenylphenylenediamine -hiline Phenybl-naphthylamine Phenyl-2-naphthylamine 4-Llethylaniline S,S'-Dimethylaniline Diphenylamine Dibenzylamine Benzylamine 1-Naphthylamine Benzidine Diphenylbenzidine p-dminodimethylaniline -I--(1-naphthy1)ethylenediamine Dioct yldiphenylamine
Original Color after Furfural Addjtion Color 2 minutes 30 minutes 90 minutes Blue Green Blue Green blue Sone Sone Sone Sone Sone Sone rone
Yellow Pink Yellow Sone Sone Pink Sone
Sone Kone Sone Orange Ppt.
Orange yellow Carmine Orange yellonOrange Sone Pinker Sone
Sone
Orange bron-n
\-ellomOrange Orange Orange pink Red orange Red orange l-ellon-
Sone Orange Pink ?;one
Sone Yellon-
Tell ow
Orange Magenta Orange Orange Trace Pink purple Kone Sone Brown Yellow Orange Pink Red orange Yellow
VOL. 30, NO. 12, DECEMBER 1958
Relative Intensity
+++ +++ +++ +++ ++++++ +++ ++ +++ ++ +
2003
Table II.
Manutacturei-a Designation 8515b
mary aromatic amines R ith furfural may involve another mechanism ( 2 ) . The most ridel? accepted mechanism ( 1 , 4 ) of the antioxidant properties of diarylaniines is
Selected Determinations of p,p’-Dioctyldiphenylarnine
Sew Gr Used Xew Se1%-
D O D I Found, 5
DOD.l Added, yL 2 00
-1IlalJ st 1
hnal>zt 2 2 0s 1 80
2 06 1 96.1 99
2 00
ArJH
+ RO?.
(rir2S). Lsedc Usedc
Sd
8200e
Lubricating oil
Peed Used Used KenSew Used Used Used Used S ew’ Used
*
c e
0.M
0.35 0.2i
0.43
0.15
1.70 1.91 2.03 1.96 1.64 1.52 1.45 1.63 0 25 0 14 0 08
2.00 2.00
0.25
Used a
0.79
1.60
1.90 1.9s 1 0“ 1. G : ;
0 24
0 14 0 03
Oronite Chemical Co. Samples consist of a specific disiloxane, silicone thickener, and diester of a dibasic acid, together nith DODh and auinizarin additives. Osdized in the laboratory.‘ Same as 8515 without quinizarin Same as 8515 without diester.
twice as much color was obtained when twice the recommended amount of furfural reagent \vas added. On the other hand, about half the color was obtained with half as much furfural. This emphasizes again the need to run standards under the same conditions as the sample. Spectra of a number of samples are given in Figure 4. The close similarity between the sample spectra and those of standards in alcohol shows that the same colored species is measured in each case. The use of strong sulfuric acid creates a difficult problem, as it causes a haze with hydraulic fluids that contain derivatives of silicon. The effect is greatest with highly oxidized samples, and slight or nonexistent with new fluids. The solid material, which is presumably colloidal silica or an insoluble silicate, is extremely difficult to remove. Centrifugation is not effective, and filters fine enough to yield a clear filtrate clog immediately. Several grades of Celite were tried a t different concentrations; KO. 503 and KO.535 make it possible either to centrifuge or to filter through glass wool. The use of Celite is an important factor in the success of this method. Test results on known fluids and on standards n-here no haze w s originally present were not changed by the use of Celite. Glass wool is preferred, as erratic results may be obtained if filter paper is used. The aqueous sulfuric acid is probably preferentially absorbed b y the
2004
ANALYTICAL CHEMISTRY
8515 USED
8200 N E W 8230 L S E D
I200
SiAhKS
SAV’LES
--f
(-1i-J).
+ R02H
stable product
The stable product formed does not interfere in this test because good correlation is obtained between the test results and laboratory oxidation studies and because, as mentioned earlier, the spectrum of the colored product on samples is identical n i t h that of unoxidized standards. A variation of color, both in hue and intensity, with structure can be expected as indicated in Table I. Because the reaction of furfural with amines is a general one. interference will occur if certain amines other than DODA are present. Under the conditions of this test, phenyl-l-naphthylamine gives a red color, pheny1-2naphthylamine an orange color, and S,.V’-diphenylphenylenediamine a blue color. The presence of these amines may be detected by measuring the absorbance a t 450 mp, where the DODAfurfural reaction product has almost no absorption. I n routine use, interference is unlikely; if present, it will usually be apparent immediately. Table I1 give. some typical results obtained from application of the test to neIv and uqed fluids. representing both field samples and those ovidized in the laboratory. Table I1 also gives similar results on hydrocarbon fluids. The average deviation is i 2% of the amount present in the 1 to 2% range and + 5% in the 0.2 to 1% range. LITERATURE CITED
I 350
300
430
470
510
550
WAVE LENGTH I N MILLIMICRONS
Figure 4. Spectra of hydraulic fluids after color development
paper, making it difficult to control the factors of acid strength and time. The filtration problem vanishes n hen the test is applied to hydrocarbon fluids rather than t o derivatives of silicon. The mechanism of the furfural-amine reaction is not known. The qualitative studies discussed earlier indicated that aromatic amines not blocked in the para position react much faster than DODA. This suggests that some aromatic ring addition or substitution is involved. However, the reaction of pri-
(1) Bickel. -4. F., Iiooyman, E. C., J . Chem. SOC.1956, 2215-21. (2) Burchfield. H. P.. Judy. J. S . . Ax.4~. CHERI.19. 786-9 11947). (3) Dyinlop,’ A . P..‘Peters, F. X., “The Furans,” AC9 Monograph 119, pp. 308, 669, Reinhold, Sew T o r k , 1953. (4)Harle, 0 L., Thomas, J. R., J . Am. Chon. SOC.79, 2943-4 (1957). RECEIVED for review February 20, 1958. Accepted June 30, 1958.
Correction I n the article on “Titrimetric Determination of Unsaturation by Catalytic Hydrogenation” [ ITilliam Seaman, ASAL. CHEM.30, 1840 (1958)],the third line in the column should read: teniperature, and nxygen is introduced.