The rate of oxygen absorption by moderately and highly refined oils can be used as a method of determining resistance to oxidation. The rate of the oxidation can be represented accurately by a simple empirical rate law. The temperature coefficient of the oxidation reaction has been determined and shown to be an important characteristic of the oil. Certain products of the oxidation (peroxides, acids, and water) follow the same general rate law as the oxygen absorption. The results indicate that the oxidation of white oils is an autocatalytic process involving a chain mechanism whose rate determining step is preceded by a rapid diss o c i a t i o n equiliRALPH W. DORNTE brium. General Electric Company, Schenectedy, N. Y.
OXIDATION
OF
WHITE OILS
.. . . .. . .
T
HE numerous methods for evaluating the
chemical mechanism of oxidation, but the determination of any one component (peroxide, alcohol, aldehyde, ketone, or acid) is insufficient for determining the course of the reaction since the chosen component may be the product of a secondary process. Unpublished results obtained by C . Van Brunt of this laboratory have indicated the importance of this aspect of the problem. If the oil is always saturated with oxygen, the rate of oxygen absorption is easily measured and unambiguous.
oxidizability of oils (1, 3) have been developed for special applications of the material but have given only comparative values which are often sufficient. Oxidation tests, designed to simulate service conditions, are difficult to interpret and aid meagerly in understanding the complex physical and chemical processes involved. Taylor (7) has stressed the importance of kinetic investigations of problems involving the resistance of materials to deterioration. The present work was undertaken to obtain data which would aid in understanding the chemical mechanism of oil oxidation. For determining the rate of reaction between oxygen and oil, diverse chemical and physical properties of the system have been used. The rate of oxygen absorption seemed to be the logical quantity to follow in orienting experiments. Data on the change of concentration of hydrocarbon compounds containing oxygen are necessary for deriving a complete
Experimental Procedure Two experimental procedures were used. One measured the rate of oxygen absorption at constant pressure by a definite weight of oil and the other permitted the withdrawal of samples for analysis during the course of the reaction : The oxygen absorption was determined in a circulatory system shown in Figure 1. The oxygen was circulated in the system by the sylphon pump which is connected directly to a geared motor (General Electric Company, 250 volt, d. e., 148r. p. m.). The two mercury traps were the most satisfactory valves for this pump. The oxygen was circulated at a constant rate of 2 liters perpinUte; the gas flow could be varied by a rheostat in series with the motor armature although the oxygen absorption was independent of the rate of circulation. The absorption cell containing 10 to 15 grams of oil was mounted in the thermostat (temperature controlled to *0.2" C.). Intimate contact between the oil and gas stream was effected in the absorption cell where a fine fritted-glass filter at the bottom dispersed the gas stream into very fine bubbles. The condenser directly above the absorption cell returned the distillate during the evacuation of the system. Water and small amounts of carbon dioxide formed during the oxidation reaction were removed from the gas stream by activated alumina and soda lime. The volume of oxygen absorbed was measured in the mercury-filled gas buret. This measurement was made by stopping the pump at the top of its stroke, equalizing pressure in the apparatus, and finally bringing the pressure in the system to that of the atmosphere by means of the leveling bulb and small oil manometer. The circulation was interrupted for less than a minute to take this reading. The experimental procedure consisted of weighing the oil sample into the absorption cell which was then sealed to the system. The system was evacuated and the thermostat was brought t o the temperature of the run. This procedure prevented initial oxidation of the sample and established the zero of the time axis at the admission of the oxygen. Oxygen was added when required during the course of the run.
FIGURE1. CIRCULATORY SYSTEM FOR OXYGEN ABSORPTION
26
JANUARY, 1936
INDUSTRIAL AND ENGINEERING CHEMISTRY
The second procedure was essentially a flow system using a larger absorption cell of the same type and same ratio of oxygen flow to the weight of oil. T h e p u r p o s e of these experiments w a s t o determine roughly the distribution of the oxygen after absorption and to o b t a i n a n oxygen balance, if possible. T h e analysis of the oil samples, withdrawn periodically, w a s for functional groups containing oxygen--1. e., hydroxyl, carbonyl of aldehyde and ketones, carboxyl and peroxide as well as u n s a t u r a t i o n . Quantitative determinations of these radicals are not of high precision a t low concentration in oils where many complications may arise. The methods of analysis have not been checked for their accuracy on oils but are believed to suffice for the present purpose. The hydroxyl group was determined by means of acetyl chloride and pyridine (6). The total carbonyl group of aldehydes and ketones was determined together by the method developed by Strache (6). The carboxyl group was titrated with potassium hydroxide according to the standard method of determining acid numbers of oils. The peroxides were determined by the rocedure of Yule and Wilson (81,which has been found satisFactory for gasoline. The unsaturation was determined by the bromide-bromate titration (2, 4 ) . The water and carbon dioxide were determined on a modification of the circulating system. A trap packed with glass wool and cooled by liquid oxygen condensed these products from the circulated oxygen. Two of these traps were arranged so that either could be connected in the circulating system. The water and carbon dioxide were displaced from these traps after warming to room temperature by a stream of purified oxygen. The products were weighed after absorption by drierite and ascarite. The results of all these determinations were calculated as equivalent cc. of oxygen at normal temperature and pressure per 100 grams of oil.
27
white oils and indicate an autocatalytic reaction. It has been found that these curves can be represented by the equation: V I / * = kt n (1) where V = total vol. of oxygen absorbed (at N. T. P., cc./lOO grams of oil) 1 = time, hr. k = the constant which characterizes rate of oxidation reaction
+
The slope of the line obtained by plotting V i / * against t gives the value of k . A series of runs is shown in Figure 3 where the data are plotted in this manner. It is evident from these graphs that Equation 1 represents the experimental data adequately and gives a cpncise method of presentation. This empirical equation has been found to represent the oxygen absorption data over the entire range of temperature and total absorption which were possible t o investigate. Except for minor deviations in early stages of the reaction, the agreement is excellent even when 10 per cent by weight of oxygen has been absorbed. The values of k for various experimental conditions and treatment of oil A are as follows:
a
Run
Temp.
21 20 19 17 34 67 27 28 22 23 18 24 36
95 105 116 125 135 145 125 125 115 115 125 126 135
IC
c.
On Pressure
Cn. 0.545 1.85 3.51 9.15 19.2 50.0 8.1.5a 8.15 2.81 1.48 6.54 3.12 21.7
76
76 (new sample) 76 (new sample) 15.2 (NI, 60.8) 2 (Nz, 74) 15.2 (Na, 60.8) 2.2 ( N I , 73.8) 76 (sample treated with 10% fuller's earth)
Glass beads.
Results &ith Oil A Most of the results t o be reported were obtained for two different samples of oil A whose properties are as follows: Specific gravity (15.5"C.) Flash point C. Fire point C. Saybolt Uhversal viscosity Acid No. Saponification No. Pour teat, C. Color
(looo F.), sec.
0.870-0.875 170 190 100-106
n 0
-25 Water-white
This oil is a typical highly refined white oil, yet the kinetics of its oxidation is characteristic of a large number of materials. The oxygen absorption by certain yellow oils which have been only moderately refined indicate the same oxidation mechanism. The data for run 17 a t 125"C. are as follows; the total volume of oxygen absorbed is calculated to normal conditions of temperature (0' C.) and pressure (76 cm.) for 100 grams of oil: Time, Hr. 0.60 1.00 1.50 2.00
Ca. 0s per 100 Grams 36 110 225 390
Time,
Hr. 2.50 3.00 3.50
c c . Or per 100 Grams 590 870 1222
Time, Hr. 4.00 4.25 4.50
cc. 0, per 100 Grams 1817 2170 2550
These data together with those a t 105" C. and 115" C. are shown in Figure 2. These curves are characteristic of the %
ROOTOF TOTAL OXYGENABAORPTIOX FIGURE3. SQUARE AS FUXCTIOX OF TIME
VOL. 28, NO. 1
INDUSTRIAL AND ENGINEERING CHEMISTRY
28
Discussion of Results The values of k for oil A at an oxygen pressure of 76 em. can be represented by the Arrhenius equation : log k = a
- Q/4.58T
where a = 14.020 Q = 23,800
-k
This is evidenced by the straight line in Figure 4 where the logarithms of k are plotted against the reciprocals of the absolute temperature. A simplified calculation of the probable error in the net activation energy Q gives *950 calories.
.ooc
.ooes
FIGURE5. EFFECTO F OXYGEN PRESSUREON REACTION RATE ,003
FIQURE4. TEMPERATURE COEFFICIENT OF REACTION RATE
The results of the analysis for oxygen radicals in the oil and determinations of the water and carbon dioxide formed are given in Table I where the results are reported in equivalent cc. of oxygen a t normal temperature and pressure per 100 grams of oil. For this calculation it was assumed that the peroxide contains 2 atoms of oxygen per molecule. The unsaturation is calculated to the same basis by the arbitrary assumption that one molecule of oxygen is equivalent to a double bond. The very low concentrations of alcohols, aldehydes, and ketones make these analyses difficult, and therefore the latter have only qualitative significance. Absorption experiments with six other commercial oils of various viscosities and degrees of refining were made. The properties of these oils are as follows: Oil designation Sp. gr. (15.5' C.) Flash ppint; O C. Fire point, , C. Sa bolt Universal visoosity 1000 aec. Aoid b o . Iodine No. Pour test ' C. Color(N. b.A.) a Water-white.
F)
B C D E F G 0.89 0.88 0.88 0.89 0.89 0.88 165 153 I66 140 175 195 190 174 180 157 I90 215 250 195 0 0 0 0 -30a -40 W.W.
150 0 4.1 -35 w.w, 2 . 5
76 0 0 -40 1.6
80 0 0 -35 W.W.
72 0
0 -50 1.0
It was found that the behavior of oil A was typical of this group. The values of the constants for Equation 1 and the Arrhenius Equation 2 are as follows: for 02 at 76 Cm.115O C. 126' C. 135' C. 3.51 9.15 19.2 3.37 10.87 21.4 4.85 9.28 25.5 7.34 16.8 4:Kl 10.5 24.6 3.65 8.00 23.0 0.70 1.63 6.26
----k Oil A
B C D E F G
145'C. 50.0 52:6 34.0 78:l 20.0
a 14.020 18.730 15.301 15.566 15.304 17.371 22.812
4 23,800 32,200 26000 26:800 27300 29:900 41,200
There appears to be no connection between the viscosity of these oils and their rate of oxidation. Oil G has an exceptional resistance to oxidation as shown by the low values of k and the high value of the net activation energy, Q. A fivefold difference in value of k for two oils under the same conditions represents a twenty-five fold difference in the total volume of the oxygen absorbed in a given time interval.
c
a
12 /e TIME CHRS)
20
M
28
z
AT Low PARFIGURE6. OXYGENABSORPTION TIAL PRESSURE
The net activation energy is not inf l u e n c e d b y c h a n g e s in the partial pressure of the oxygen. It is apparent t h a t t h e oxidation rate of this oil varies appreciably from one lot to another as s h o w n b y runs 17 and 27. The r e a c t i o n which is measured by t h e absorption of oxygen is not influenced by a fivefold increase in the glass surface in c o n t a c t with the oil and oxygen as shown by runs 27 and
40
.IU A
s
4 30
&E >
2 c,
EO
IO
FIGURE7. EFFECTOF OILTREATMENT ON RATEOF OXIDATION
JANUARY, 1936
INDUSTRIAL AND ENGINEERING CHEMISTRY
29
tered. This treatment was evidenced in the oxidation data by a 10 per cent increase in k and a negative value of n as shown in Figure 7. The value of n is probably related to the inhibition of the reaction by unavoidable foreign COz 0 materials and natural inhibitors in the oil. These inhibitory effects occur in many reactions whose 0 chemistry is well known, and are either very 0 0 difficult or impossible to control and reproduce. 1.9 In the present work it seems justifiable to neglect 7.3 this small correction term in Equation 1 which 14 2 then becomes : Vll2 ht (3) Very large negative values of n are obtained by adding minute amounts of phenyl-a-naphthylamine. A concentration of 1 part ip 10,000 causes an inhibition period of about 20 hours, during which no absorption of oxygen can be measured. Figure 8 illustrates the remarkable inhibitory effects of this material. After the inhibition period the oxidation reaction proceeds a t nearly the same rate as the uninhibited oil. After this oil has absorbed a few tenths per cent of oxygen, the addition of phenyl-a-naphthylamine a t 0.5 per cent concentration produces no inhibition period and only a slight decrease in the oxidation rate. The changes in composition of the oil during oxidation are illustrated by the data on oil A (Table I). These results are not all of the same order of precision. It is believed that the determination of peroxide, acid, water, and carbon dioxide are much more reliable than the other determinations. This is due to limitations of the methods and the low concentrations of the alcohols, aldehydes, and ketones, as well as unsaturated materials. The former group of products conforms to the general rate law (Equation 1) as shown by Figure 9. The straight line representing the water formation coincides with that of the acid produced; even the small deviations a t the beginning of the reaction are the same for these two products. The slopes of the lines representing the total oxygen absorbed, peroxide, water, and acid have the respective values: 19.2, 10.4, 5.86, which are approximately in the ratio of 4:2 : l : l . At any time approximately 25 per cent of the total oxygen absorbed is found as peroxide, 6 per cent in the water, and another 6 per cent in the acids produced. The carbon dioxide formation was negligible and carbon monoxide was not detected b s analysis of the gas. The other products of oxidation do not follow the general rate law and may r e s u l t from side reactions. The oxygen abs o r p t i o n experiments clearly demonstrate t h e autocatalytic nature of the oxidation. The inhibition periods produced by minute concentration o f an antioxidant indicate a chain mechanism of considerable l e n g t h . 2 3 4 From Eauation 1 FIQURE 9. DISTRIBUTION OF OXYGEN the rate Of the reaction is : AFTER ABSORPTION
ANALYSISFOR OXYGEN RADICALS IN OIL A AND OF WATER AND CARBON DIOXIDE FORMED (135' C., OXYQENAT 76 CM.)
TABLE I. Time, Hr. 0 0.17 0.33 0.50 0.75 0.92 1.17 1.50 1.89 2.42 2.84
Peroxide 0
(In cc. of oxygen at N. T.P. per 100 grams of oil) CarHy- Unsatu- Time, bony1 droxyl rstion Hr. Hz0 Acid 0 0 0 0 0.75 11.5
11 17.5 55.3 101 126 217
.. .
386 590 1010
3:7 9:s 7.9 17.0 34.5 65.9 294'
... ...
... 138
84 149 163 107 259
...
..
..
.. 48
.0.
8
..
58
.. 56 .. ..
0
49 66 61 63
1.00 1.42 1.75 2.25 2.75 3.25 3.83 4.25
E
53.5 107 159 255 406 515
13i
:
E
28. Preliminary experiments showed that the oxygen absorption rate was decreased slightly by the water and carbon dioxide formed during the process. These products were removed continuously in all the runs reported. The values of k vary but little for large changes in the partial pressure of oxygen in the circulated gas. The values of k a t constant temperature vary approximately with the onefourth power of the partial pressure of oxygen as shown in Figure 5. Preliminary experiments have shown that nitrogen has no inhibiting effect on the rate of oxygen absorption. There is no deviation from the general rate law (Equation 1) even when the oxygen partial pressure is reduced to 2 cm. by dilution with nitrogen. The agreement between Equation 1 and the data is borne out by Figure 6. The catalytic compounds formed by the reaction are quite stable under the experimental conditions as shown by run 23 in Figure 6. It is difficult to assign physical significance to constant n in Equation 1. To discover the effect of oil treatment upon the values of n, oil A was treated a t room temperature with 10 per cent by weight of fuller's earth, centrifuged, and fil-
FIQURE 8. INHIBITION BY VARIOUS CONCBNTRATIONB OF
PFCBNY~~U-NAPHTHYLAMINE
VOL. 28, NO. 1
INDUSTRIAL AND ENGINEERING CHEMISTRY
30
dV/& = 2kV1’3
(4)
where V represents the total volume of oxygen absorbed and is presumably proportional to the concentration of an unknown reactive substance, AB,in the oil. Substance A Pmay contain oxygen although this is not a necessary condition for the empirical rate law. If substance A? does exist, a mechanism of the type,
-
(AI2 2A K = (A,) Rate-determining step A ?
Rapid equilibrium A2
+
may be suggested to explain the empirically established rate law, Equation 4. It is evident from this mechanism that the rate of reaction will be proportional to the square root of the total oxygen absorbed. The results available do not justify a detailed discussion of the kinetics and scarcely suffice to prove that the foregoing mechanism is correct even as to type.
I n acknowledgment, the writer wishes to express his appreciation of the suggestions and criticisms of A. L. Marshall and associates of this laboratory and of the assistance of Earl T . Marx in the experimental work.
Literature Cited (1) Ellis, “Chemistry of Petroleum Derivatives,” New York, Chemical Catalog Co., 1934. (2j Francis, IND.ENG.CHEM.,18, 821 (1926). (3) Kalichevsky and Stagner, “Chemical Refining of Petroleum,” New York, Chemical Catalog Co., 1933. (4) Mulliken and Wakeman, IND. ENG.CHEM., Anal. Ed., 7, 59 (1935). (5) Smith and Bryant, J . Am. Chem. Sac., 57,61 (1935). (6) Strache, Monabh., 12,524 (1891); 13,299 (1892). (7)Taylor, Proc. Am. SOC. Testing Materials. 32, Part 11, 1-34 (1932). (8) Yule and Wilson, IND.ENQ.CHPM..23, 1254 (1931). RECmIVBD June 8,1935. Preaented before the Division of Petroleum Chemistry at the 89th Meeting of the Amerioan Chemioal Society, New York,
N. Y , .4pril22to 26, 1935.
Effect of Mixed Acids
UPON IRONS AND STEELS JUSTICE EDDY AND F. A. ROHRMAN Michigan College of Mining and Technology, Houghton, Mich.
High-carbon steels show a greater tendency to retain their passivity in mixed acids than low-carbon steels. Quenched steels show a greater tendency to resist mixed acid than furnace-cooled steels.
S
ULFURIC-NITRIC acid mixtures have tremendous importance in the manufacture of nitrated cellulose, benzene, and glycerol. The percentage of mixture depends upon the type of nitration and process, but usually contains from 20 to 50 per cent nitric acid (by volume) with small amounts of water. Fortunately, because of their remarkable passivating property, sulfuric-nitric acid mixtures can be handled and transported with iron or steel equipment. It occurred to the writers that some steels and irons might be more easily passivated than others and hence be more resistant to corrosion; consequently an investigation was carried out with the view of studying the effect of mixed acids upon various steels and irons under different conditions.
Experimental Procedure Mixed acids of varying acid and water contents were employed a t different temperatures. Special care was taken to obtain standard S. A. E. rated steels which shOwed very little differences in their phosphorus, sulfur, silicon, and manganese contents, These steels were all given the same heat
treatment in order to eliminate any differences due to strains and fabrication. All the samples were size‘d to 2 X 2 X ‘I8 inch (5 X 5 X 0.32 cm.), polished, cleaned, and weighed. They were then exposed to the mixed acid solution for 100 to 1000 hours in 600-cc. beakers containing 400 cc. of mixture; finally they were washed with water, dried, and reweighed. The losses in weight were calculated on the basis of loss in grams per square decimeter per 100 hours.
Effect of Carbon Content All the classical experiments involving the corrosion of iron by dilute solutions show definitely that carbon acts as a corrosion accelerator. It has been demonstrated that iron freed of carbon and other more electropositive impurities has very little tendency to corrode. The tests in mixed acid show an opposite tendency, however. The following table gives the results after exposure to a mixture of 40 per cent sulfuric acid, 40 per cent nitric acid, and 20 per cent water: Carbon Per cent 0.95 0.87 0.55 0.35
Loss i n 100 Hr. at 25O C.
Carbon
Gram/sq. dm.
Per cent
0,2952
0,20 0.08 0,02
0.3524 0.3568 0,3645
Loss i n 100 Hr.
at 25’ C. Gram/sq. dm. 0.5073 0.5202
(Constant Hz evolution)
The results indicate, without doubt, that the steels high in carbon are the more resistant. A number of irons with carbon contents ranging up to 4.5 per cent were also exposed to the same acid and under the same conditions. I n every case the irons stood up as well as the S. A. E. 1095 steels. Inas-