March 1954
INDUSTRIAL AND ENGINEERING CHEMISTRY
advantageous. The same applies to a lesser degree to ester groupings, one ester grouping per molecule being more desirable than two. These deductions apply, however, to compounds of a certain molecular weight range. If this range is greatly exceeded, the above statements may no longer be true. The N-ethyl-AT8-hydroxyethylbenzenesulfonamide esters show up somewhat better than the corresponding N-butyl derivatives of the same molecular weight. It would be of interest to compare them with iY-methyl and AT-propyl derivatives. The oleate ester in Series I1 is somewhat better than the stearate in per cent elongation and is far superior to it in minimum flex. It euggests that the dodecenate may be far superior to the laurate. The homologous series of Series I1 and Series I11 shows that unfortunately maximum values for elongation do not correspond to minimum values in minimum flex temperature and that a compromise choice may have to be made. It is evident that when phenyl is substituted for an alkyl in .V-alkyl-Ar-(3-hydroxyethylbenzenesulfonamideacetate, much poorer elongation and much poorer minimum flex result (Series IV). Comparing compounds of approximately the same molecular weight in Series VII, i t is evident that N,Ar-di-n-octyl-p-tduenesulfonamide is a much better plasticizer than ilT,N-diethyldodecylbenzenesulfonamidp. N,N-Dialkyl Tetralin or naphthalenesulfonamides make poor plasticizers (Series VI1 and VIII). Thiophene derivatives Ehow promise, but not enough individual compounds have been tested to show that they have a real advantage over the cheaper benzene derivatives. Kuclear chlorination makes a t least the lower molecular weight members of N,.V-dialkyl aryl sulfonamides incompatible with the Vinylite resins. N,Y-Dialkenyl compounds are apparently no more desirable than the N,N-dialkyl compounds (Series VI). The limit of compatibility of X,N-dialkyl aryl sulfonamides and the Vinylite resins is reached somewhere between h',"Vdi-n-octyl-ptoluenesulfonamide and K,N-didodecylbenzenesulfonamide, Series VI and VII. W ~ X E S .J17axes are valued primarily for their physical properties. Their chemical conqtitution i p of little interest to their
591
users. Their main users are manufacturers of paper products, polishes, electrical equipment, carbon paper, and textiles. The properties desired thus vary with the use to which a wax is put. The usefulness of the waxlike materials prepared was examined only a i t h respect to polishes. Waxes are used in polishes chiefly as wax-in-water emulsions or as gels of wax in solvents such as turpentine, and mineral spirits. A11 the waxes listed, a i t h the exception of the thiophene one, gave reasonably good gels with turpentine or mineral spirits. All imparted good luster to leather, nood, and metal surfaces. \Tax digels made from Ar,K-di-p-hydroxyethylbenzenesulfonamide stearate synergized on standing a t high room temperature (30" to 50). When a gel was made from X-dodecylbenzenesulfonamide it did not produce any shine on leather, wood, or metal surfaces and destroyed shine imparted to those surfaces by waxes. However, N-octadecyl-p-toluenesulfonamide imparted good luster to wood but a poor luster to leather. The gels were generally prepared by pouring 25% solutions of the wax in turpentine or in mineral spirits at 40' to 50". Emulsions in which anionic agents, such as morpholine oleate, were used had a tendency to thicken to gel on standing. Considerable dilution a i t h LTater would not materially thin the gel. It is thought that the thickening is due to the reversion of the oiI in water to a water-in-oil system. ACKNOWLEDGMENT
The author wishes to thank D. H. Vheeler for encouragement and cooperation. Analytical data and evaluation were done under the direction of Harold Boyd. LITERATURE CITED
(1) Bergen, H.
(2) (3)
(4) (5)
S.,and Craver, J. K., IKD.ENG.CHEM.,39, 1082-7 (1947). Clash, R. F., and Berg, R. bl., Ibid., 34, 1218 (1942). Pearock, D. H., and Dutta, U. C., J . Chem. SOC.,1934,1303-5. Reed, M. C., IND.EKG,CHEM.,35, 896 (1943). Van -4ntwerpen, F. J., Ibid., 34, 68-73 (1042).
ACCEPTED December 2, 1953. RECEIVED for review August 5 , 1953. Paper 156, Journal seriea, Research Laboratories, General Milla. Inc.
Displacement of Petroleum from Sand Surfaces by Solutions of Polyoxyethylated Detergents H. K. DUNNING, H. J. GUSTAFSON, AND R. T. JOHANSEN Surface Chemistry Laboratory, Petroleum Experiment Station, Bureau of Mines, Bartlesville, Okla.
I
N RECEKT years wide attention has been given the possibility of obtaining a more complete recovery of petroleum from partly depleted reservoirs by the use of surface active additives in water-flooding operations (3,16). Since their introduction, nonionic detergents have appeared to be among the most promising types of surface active additives. The number of species of various sizes produced in the reaction of a phenol with a given mole ratio of ethylene oxide may be represented by Poisson's distribution law (11, 25). Since the mole ratio of ethylene oxide can be varied continuously above a value of I, this reaction can produce detergents of any desired composition (IO). For a given phenol, the length of the polyoxyethylene chain determines the hydrophilic-lipophilic
-
balance of the molecule (4). Recently, complete series of pol) oxyethylated detergents have become available in research quantities. The availability of these series permits a comprehensive investigation of the effects of hydrophilic-lipophilic balance on the displacement of petroleum from solid surfaces. The efficiencies with which these detergent solutions displace petroleum from hydrophobic sand surfaces has been investigated by a centrifugal displacement method. The ability of a detergent solution to displace petroleum from reservoir surfaces is a primary requirement if the detergent is to aid in the more complete recovery of petroleum from a reservoir that has hydrophobic surfaces. However, this may not be the only requirement, and a better displacement of petroleum may
592
INDUSTRIAL AND ENGINEERING CHEMISTRY
not necessarily result in more complete petroleum recovery since drops of oil, although displaced from reservoir surfaces, may he lodged in the irregular openings of the porous medium.
TABLE 11. 40.
M4TERIALS
SY\THETIC OIL S 4 \ D f . Synthetic oil sands were piepared bv mixing 2 Irg. of Railioad I T hite sand (40 to 70 mesh) n i t h 136 2 grams of the crude oil samples. This was about the maximum amount of crude oil that could be used n-ithout excessive gravity drainage. The surface area of the sand, as determined by microscopic methods, was 112 square cm. per gram.
DETERGESTS
Vol. 46, No. 3
INVESTIGATED
of
Series Samples Description I 7 Products of continuous oxyethylation process of Hercules Powder Co. I1 Products of h n t a r a Chemicals Division, General Dyestuff
Range of Ethylene Oxide-Phenol Phenol (Mole Ratio, E ) Alkyl 4-1 to 20-1 Nonyl
1 5-1 t o 3 0 - 1
Corp.
I11 IV
i
Products of Rohm &- Haas Co.
5
Dialkylphenylpolyoxyethylene ethers produced by Oronite
Octyl 1-1 t o 40-1 Dialkyln 30-1 to 210-1
Chemical Co.
.iv. no. alkyl carbon atoms
Figure 1.
Oil Displacement by Centrifugal Method
'The sand initially was cleaned with hot chromic acid, washed thoroughly with tap water, distilled n-ater, and acetone, and dried a t 110" C. The crude oils were topped a t 50" C. to renioyc the highly volatile components. This treatment minimized erratic results caused by evaporation during the determination and caused only minor changes in the specific gravities and viscosities of the crude oils. Samples of the oil sands were removed from the container only after thorough mixing and betwen tests the oil sands were agitated intermittently. P E T R O L E U M SAMPLES.Crude oils from the Rio Bravo field in Kern County, Calif., and the Oklahoma City field, Oklahoma County, Okla., were used in preparing the synthetic oil sands. The sources of the crude oil samples were described recently (6, 7). Some of the properties of thcse crude oil samples arc summarized in Table I.
TABLE
I.
I'R 3PERTIES OF ~ l ? T R O L E U . \ fSAMPLES
Viscosity ____-.___ Spec. Grai-ity API 0.828 3A 4
____-.
6 O j 6 O 0 F.
Rio Bravooil Oklahoma City Wilcox oil
0.834
" Oil-mater interface
38.3
2.68
Sayholt, see. a t 2j0 C. 37.4
S.42
48.8
Ci, a t
2 j 0 C.
Interfacial Tensiono, Dynes/ C m . at 2 5 O C.
19.8
19.9
aitPr 1 hoai
The values of interfacial tension a t the crude oil-water interface were determined with the pendent-drop instrument (1, 8). This dcgree of interfacial activity indicates considerable contents of polar constituents in these petroleum samples. It mould be expected that these constituents would promote the wetting of solid surfaces by the oily phase. There is considerable evidence
=
29.
that petroleum samples from t>hesefields cause clean sand surfaces to become hydrophobic. Visual observations of the capillary rise of these oils in clean sand columns gave substantiating evidence. Therefore, the results of these studies should be considered applicable primarily to systems involving hydrophobic surfaces and not to systems involving hydrophilic surfaces. DETERGESTG.The detergcnt,s used in these studies were of thc polyoxyethylated alkyl phenol type. Each manufacturclr furnished a series of detergents in which the lipophilic group (alkyl phcnol) remained constant while the hydrophilic chain (po1yoxyeth:-lene) varied in length. These series are described in Table 11. The 24 detergents were used in the form in which they wwc~ received from the manufacturers (95 to 100% active). Although the addition of inorganic builders may increase the petroleumdisplacement efficiencies of these detergents or decrease thc amount required, the studies were limited to investigations of the effects of solutions prepared with pure detergents. I n this report the detergents are indexed by a Roman numeral representing the series and an Arabic numeral representing thc mole ratio of ethylene oxide to alkyl phenol. For example. 1-4 represents the detergent from series I having a mole ratio of 4 to 1. The term "mole ratio of ethylene oxide to alkyl phenol" is represented by R. EXPERIMENTAL
OIL-DIsPr,.%cExEsT EFFICIESCIES. Thirty granis of a synthetic oil sand were placed in a centrifuge tube having a calibrated neck of small diameter. The tube then was filled to a reference mark high on the neck tvith the detergent solution to be test,ed (about 30 ml. of iolut,ion). Triplicate tests mere made on groups of eight, in which one of the tubes contained distilled n7atc.r as a standard. The tubes iiiit,iallywerecentrifuged for 16 minutes. They were removed from the centrifuge, tilted to about 45". swirled gently to allow release of trapped oil drops, and centrifuged for another 10 minut,es. This treatment was repeateJ four t,imcs. After each centrifuging the amount of oil di+ plared from the sand was measured directly in the calibrate I neck. Although it was observed that the oil displa-eJ remairie 1 constant after about two treatments,, the entire procedure waq adopted to ensure the attainment of equilibrium. An International centrifuge, size 1, Model SB,operated at, 2500 r.p.m., vas used in these test.. Under the-e condit,ions the relative centrifugal force 'Ivas 1200 and the pressure differenm between the t v o phases was 69 em. of mercury a t the center of the sand sample. The application of the centrifugc to the k s t ing of petroleum reservoir materials has been tiemibed b y Slobod et al. ( 1 7 ) . SURFACE TESSIOSS.Surface tension value. rvere determined by the du S o u v ring method. A converted Chainomatic balance \?,-as used for the measurements. A platinum ring of 4.0-em. radius mas employed and the instrument was damped slightly for ease of operation. Although several corrections must be applied if abFolute values of surface tension are desired (19, 15, 19), it x a s deemed sufficient to incorporate them into one factor by measuring the surface tensions of pure liquids and arriving a t an empirical correct,ion factor. Accordingly, the instrument
March 1954
I N D U S T R I A L AND E N G I N E E R I N G CHEMISTRY
was calibrated over the range of surface tension to be investigated by measuring the surface tensions of purified benzene and nitrobenzene. In this range y (corr.) = y (obsvd.) X 0.93
where y is surface tension in dynes per cm. Temperature and level control were obtained by using a combination thermostatelevating platform (9).
EXULSIFYING CAPACITIES.An attempt to standardize several detergent tests is described by Harris (14). However, the emulsifying capacity of a detergent depends on many factors, most of them unknown, as well as on the substance being emulsified. Accordingly, various empirical methods may be used to measure this property (16). In order to evaluate the relationship between emulsifying capacity and crude oil displacement, an empirical test was established to determine the emulsifying capacities of the detergent solutions with the Oklahoma City Wilcox crude oil. Ten milliliters of detergent solution were placed in a 30-ml. graduated test tube. Two milliliters of crude oil were allowed to flow gently down the side of the tube. Then the tube was tightly stoppered and placed in an extension arm of a Burrell wrist-action shaker, shaken for 3 minutes, and placed in a vertical position, and the time was recorded a t which a separation of the oil and detergent solution became visible. This end point was sharp and easily determined except for a few of the oil-soluble detergents. Operating conditions must be duplicated meticulously if the results in such an empirical test are to be duplicated. The test tubes were petroleum distillation tubes graduated in 0.2ml. divisions from 0 to 30 ml., as described by Dean et al. (6). The shaker was timed a t 335 cycles per minute. The center of the sample was located a distance of 23 cm. horizontally from the center of the shaft, a t a 45' angle with the vertical. The amplitude of displacement was 9 em. The amount of agitation selected was sufficient to give a measurable time of phase separation for the weaker emulsifiers and a reasonable time for the best emulsifiers studied. CLOUDPOINTS.Cloud points for 1% detergent solutions were determined with a TAG thermometer calibrated a t 0' and 25" C. The cloud points were very sharp as reported by Fineman and others (4, 10). A few of the cloud point determinations were repeated with 0.1 weight % solutions. Only slight differences were observed between the cloud points for the 1% and 0.1% solutions, as noted in previous reports (4, 10). REFRACTIVE INDICES. Refractive indices of the detergent solutions were determined with a Precision Abbe refractometer a t 20" C. Detergent solutions with a cloud point below 20" C. were agitated violently and the refractive indices were determined immediately, Although the field division was blurred with some of the dispersions, the refractive index values could be reproduced to within about =I=O.OOOl. RESULTS AND DISCUSSION
DISPL.4CEMENT EFFICIENCIES.The term "displacement efficiency" is defined as the amount of oil produced from a synthetic oil sand sample by a detergent solution relative to the amount displaced from corresponding samples by water. Displacement efficiencies of the four detergent series were determined with 0.1 weight % solutions. At least six determinations were made for all detergents, with the exception of IV-30 and IV-90, for which three determinations were made. As many as 15 determinations are represented in the average values for several of the detergents. Results of this type of test are subject to random errors in the sampling of the sand. No way to eliminate these errors entirely has been found. Therefore, it is essential that conclusions be drawn from statistical calculations rather than from the observation of a single or few tests. Statistical methods de-
593
scribed by Youden (18) were used in the evaluation of finite sets of data. Visual examination of the synthetic oil sand after a centrifugal displacement test gave convincing evidence of the ability of several of the detergents to remove oil from sand. After being treated with detergents of high calculated displacement efficiency, the sand was nearly white and flowed freely. Samples of the oil sand after tests using distilled water remained brown and formed a coherent mass in the bottom of the tube. The most oil-soluble detergents of series I, 11, and 111 produced about the same effect as water. Figure 1 is a photograph of four synthetic oil sand samples after the centrifugal displacement test with (1) 0.1 weight % solution of 111-5, (2) 0.1 weight % solution of 111-12.5, (3) distilled water, and (4)0.1 weight % solution of 111-1. The average amounts of Oklahoma City Wilcox and Rio Bravo crude oils displaced from 30-gram samples of the synthetic oil sands by water are shown in the following table.
Oklahoma City Wilcox Rio Bravo
Oil Displaced, MI. 1 .E5 1.47
S!d. Deviation 0.099 0.219
Per Cent of Total Oil 68.3 64.7
Apparently the lower standard deviation for the Oklahoma City Wilcox samples is due to improvement in experimental techniques, since the Rio Bravo samples were tested early in the study, The t-test (18) shows that the probability of this slight difference in displacement being valid is 85%. The maintenance and proper sampling of a synthetic oil sand are difficult. Despite precautions to ensure uniform sampling, a slight constant increase in oil displacement by water was observed as the sand was successively removed from the container. This effect appeared to depend on the depletion of the Pynthetic oil sand rather than on the time elapsed before obtaining the sample. As an example, the average oil displacement from three samples taken from the filled container was 1.52 ml., while from three samples obtained shortly before the batch of oil sand had been depleted it was 1.64 ml. The standard deviations %-ere0.186 and 0.060, respectively. The probability that this increase is valid is 98%. This effect of gradual enrichment of the oil sand is minimized by calculating the displacement efficiencies of the detergents relative to the average of three water-displacement values determined on sand samples removed from the container a t the same time as the samples used for the detergents. Considering the average relative displacement efficiencies, series I, 11, and 111 were dightly more efficient in displacing oil from the Oklahoma City Wilcox synthetic sand than from the Rio Bravo synthetic sand. The over-all average displacement efficiency for these three series was 1.25 and 1.23 for the Oklahoma City Wilcox and Rio Bravo sand samples, respectively. The corresponding standard deviations were 0.064 and 0.065. The t-test shows that the probability of this small difference being valid is 60%. The results of all the water-displacement tests and detergent tests indicate that Oklahoma City Wilcox crude oil is slightly easier to displace from sand than is the Rio Bravo crude oil. The average standard deviations for series I, 11, 111, and IV are 0.047, 0.075, 0.070, and 0.134, respectively. The increase in the standard deviation of series IV is attributed to the fewer number of determinations made with this series. The displacement efficiencies of the four detergent series in 0.1 weight % solutions are shown graphically in Figure 2 . Series I and I11 show maxima in displacement efficiency where R equals 6 and 5 , respectively. For detergents of these series having R values above 12, the efficiency is relatively constant. Series I1 exhibits a maximum in efficiency a t an R value of 15, while the maximum for series IV is not reached until R equals 120. The behavior of series I1 resembles that of I and 111 except that the
594
INDUSTRIAL AND ENGINEERING CHEMISTRY
maximum displacement efficiency occurs a t a higher R value. The maxima a t R values of 4 to 8 observed with series I and I11 could not be checked with series I1 because of a lack of samples between detergents 11-4.5 and 11-9.5. Series I V differs from the preceding three in that there is no sharp change in efficiency as R is varied.
Vol. 46, No. 3
additional methylene group in the hydrocarbon chain. Thv cloud points recorded in Table I11 corroborate this observation and indicate that it can be extended roughly to include detergents based on dialkyl phenols. Detergents 11-15 and IV-30 have nearly equal cloud points; detergent IV-30 represents an increase of about 15 ethylene oxide units and about 20 methylene groups over detergent 11-15, Thus, each additional ethylene oxide unit is roughly equivalent to a methylene group even when MOLE RATIO ETHYLENE OXIDE SERIES Ip the additional methylene groups arc in two alkyl chains. 0 30 60 90 120 150 180 210 The fact that corresponding members of series I and 111 having similar cloud points also have similar displacement efficiencies indicates that the hydrophilic-lipophilic balance of a detergent is one of the determining factors in oil displacement. EFFECTSOF COSCEATRZTIOY. The effects of concentration upon displacement efficiencies were determined for a representative number of detergents of each series. Most of the displacrment efficiencies were determined a t a concentration of 0.1 weight %. This concentration was selected arbitrarily, sinre it is above the critical micelle concentration. Although insufficient data n-ere available to draw conclusions as to the effect of concentration on the action of a specific detergent, some general trends were noted. The displacement efficiencies were I O " ' " " " " ' " " I " not affected markedly by decreasing the concentration from 0.1 0 4 8 12 16 20 2 4 28 32 36 40 to 0.02 weight %. A further decrease in concentration to 0.005 MOLE RATIO ETHYLENE OXIDE SERIES I,iI,€iIlI weight % was accompanied by a considerable decrease in disFigure 2. Displacement Efficiency of Detergents placement efficiency. At the lower concentrations detergents having high R values apparent11 maintained a greater comparative degree of displacement efficiency than did the deCLOUDP O I ~ T SThe . solubilities of these detergents decrease tergents having low R values. as the temperature increases and detergents having low R values are soluble only in cold water, if a t all. The solubilities of these REFRACTIVE IXDICES. The changes in refractive indiccs detergents are readily determined by observations of their caused by dissolving or suspending a detergent, to the extent of I%, in distilled water are recorded in Table IV. In general, cloud points. The cloud points of 1 weight % solutions of these solutions that have cloud points considerably below 20" C. detergents are summarized in Table 111. cause only a small increase in the refractive index of water. For solutions that have cloud points slightly below 20" C.-that is, approaching complete water solubility-the change in reTABLE 111. CLOUDPOISTSOF 1% DETERGENT SOLUTIONS fractive index generally is the same as though the detergent 1m-e Cloud Pt., Cloud Pt., dissolved rather than suspended. .4t very high R values the R c. R c. change in refractive index decreases slightly owing to a decrease SERIESI11 SERIESI in the refractive indices of the pure detergents as R is increased 1
