Physical Properties of Coal Tars - Influence of Tar-Insoluble and

Physical Properties of Coal Tars - Influence of Tar-Insoluble and ... For a more comprehensive list of citations to this article, users are encouraged...
1 downloads 0 Views 2MB Size
Physical Properties of Coal Tars Influence of Tar-Insoluble and Solvent-Insoluble Constituents OAL tars are distilled to produce refined tars and pitches which are used extensively for the maintenance and construction of roads, for roofing and waterproofing purposes, as binders in fuel briquets and carbon electrodes, and for other important industrial purposes. The suitability of a given refined tar or pitch for any particular purpose is usually judged by one or more of the following physical and chemical tests which, by experience, have been found to aid in the identification or evaluation of the material-moisture, specific gravity, consistency or viscosity, distillation, softening point of distillation residue, and total bitumen.

Status of Recent Tests I n recent years tests other than those enumerated have received some attention. Especially important in this connection are tests which have been proposed for the determination of: (1) amount and character of suspended material, (2) solvent-insoluble constituents, (3) viscosity index, (4) surface tension, and (5) adhesion to aggregates. The present status of each of these tests is discussed briefly in the following paragraphs.

SUSPENDEDMATERIAL. Nellensteijn ( 10) attaches great significance to the number of visik particles in a given volume of tar. He tested both unfiltered and filtered tars in nitrobenzene dispersions and contends that a “micron count”-i. e., number of visible particles per cu. mm. of taraids in the classification of tars and may be used as a measure of binding power. There has been much discussion as to the significance of this test with the result that it is now included in the specifications of some countries and excluded in others (13,17,18, 20). SOLVENT-INSOLUBLE MATERIAL. Hodurek (7) was one of the first investigators to filter tars without the aid of a solvent and to differentiate between the suspended material in tars and the solvent-insoluble material in filtered tars. He isolated the benzene-insoluble material from filtered tars and described it as “fusible and capable of great binding strength on solidification.” For this reason he infers that a knowledge of the amount of benzene-insoluble material in the filtered tar is useful as a n indication of binding strength. Klinkmann (8) supports this view. To use his terms, “the resinous constitutents in coal tar are solely responsible for its adhesive qualities.” I n the case of pitch used for the briquetting of coal, Demann (1) suggests that the portion in72 1

722

INDUSTRIAL AND ENGINEERING CHEMISTRY

VOL. 28, NO. 6

shows good adhesion, 0.5 gram soluble in petroleum ether be of the same mixture is boiled in used as an indication of adheWhen comparing distilled coal tars of each case with 6 cc. of sodium siveness. carbonate solutions of increasing the same consistency and type, correlaconcentration for o n e m i n u t e , VISCOSITYINDEX. Changes tion of the evidence obtained shows that until a separation takes place. in viscosity with temperature The concentration which is barely specific gravity is of primary importance. have been studied by many insufficient to bring about separaI t appears to serve somewhat the same vestigators (3, 3, 6, 8, 19) with tion i4 taken as a measure of the adhesion tension. the result that definite relationpurpose as determinations of suspended Even though this method has ships have been e s t a b l i s h e d . phase or determinations of solvent-inbeen a c c e p t e d i n t h e German Nevertheless, there is little inprocedures, it has not been insoluble constituents. formation in the literature convestigated sufficiently to permit In general, for coal tars of a given type a complete understanding of its cerning the relationship between significance. and of given viscosity, increases in specific viscosity i n d e x and suspended phase or solvent-insoluble congravity appear to be accompanied by Obviously, determinations of stituents. s u s p e n d e d material, solventincreases in amount of suspended phase SURFACETENSION.Surface insoluble materials, v i s c o s i t y and amount of solvent-insoluble contension determinations on coal index, surface tension, and adstituents. Surface tension and adhesive tars have been carried out by hesion t o a g g r e g a t e s require properties can also be estimated from Nellensteijn (11), Klinkmann further study if they are to be specific gravity determinations. How(8),and Fricke and Meyring (4). e m p l o y e d , either alone or in Nellensteijn determined the surc o n j u n c t i o n with the usual ever, these properties are not affected by face tensions of a coal tar and physical a n d c h e m i c a l tests the suspended phase. Since the sustwo other bituminous materials. e n u m e r a t e d above, for the pended phase has a significant influence He found that the surface tenevaluation of coal-tar products. on specific gravity, surface tension and Realizing the need for fundasion values of all three mateadhesive properties appear to be more rials increased gradually with mental research in this field, a n investigation was undertaken decrease in temperature until a closely related to the specific gravities which had for its purpose the definite temperature was reached of the coal tars from which the suspended application of suspended maat which the increase became phase has been removed. terial det,enninations, solvent much more rapid. Nicholson extraction tests, viscosity index, (18) concluded from this besurface tension determinations, havior that some asphaltic maand adhesion m e a s u r e m e n t s terials and coal tars have surface tensions higher than water a t ordinary temperatures, whereas to siniilar products obtained from typical. American coal tars a t elevated temperatures the opposite is the case. Based on of different types and origin and the correlation of these tests with the usual tests enumerated above. this phenomenon Nicholson devised a method which makes it possible to determine in physical units the adhesion tension Selection and Preparation of Tar Samples between bituminous materials and aggregates. The success of this method depends upon the reliability of the general characNine typical American coal tars were selected for this investigation. Two of these tars were produced in gas ter of the temperature-surface tension relationship asindicated by Nellensteijn. As expressed by Nicholson, “it remains plants using horizontal retorts and two in plants using continuous vertical retorts, and the remaining five came from to be shown, of course, whether the straight-line temperaturecoke-oven plants. The samples were selected to represent surface tension relation holds a t comparatively low tempera.the range of commercially aGailable tures.” coal tars. For instance, in the case ADHESION TO AGGREGATES.The of the coke-oven tars, sample 3-C various tests proposed for the dea p p r o a c h e s the specific gravity termination of adhesion of bitumirange of vertical-retort tars, whereas nous materials to aggregates recently sample 7-C approaches the gravity have been reviewed by Riedel and range of a horizontal-retort tar. Weber (16). I n the same publicaTable I gives a record of the type tion the author described a new test of carbonization equipment and of method which has been adopted as the operating conditions under which tentative in the German procedures the individual tars were produced. for the testing of road materials. The information was secured from This test is p e r f o r m e d as follows the coke or gas plants. I n :fer(16): ences to the individual samples, the numbers given in the first column The aggregate to be tested, having a fineness K-2 (K-2 = 100 per cent will be used with their respective through 28 mesh, 0 per cent through letters to designate the type of retort 65 mesh) is mixed intimately with the or oven. binder to be examined-for example, in the proportion of 71 per cent by volume The tars as received for the inof aggregateand 29 per cent of bitumen. v e s t i g a t i o n were crude tars with Approximately 0.5 gram of this mixmoisture contents between 0.5 and ture is placed into a test tube, water 4 per cent. Their viscosities varied is added, and the mixture is heated to boiling for one minute. If a diswithin limits which made a direct placement of the bitumen from the comparison of their properties diffiaggregate occurs, poor adhesion is incult. I n order to remove the free dicated. If this p r e l i m i n a r y t e s t 1

JUNE, 1936

INDUSTRIAL AND ENGINEERING CHEMISTRY

and the washings were rejected. The filter shell was then removed from the filter cylinder and placed in an air-tight container for vacuum evaporation. The evaporation was carried out over a period of a t least 12 hours in a drying oven set to maintain a temperature of 175°C. The average absolute pressure, maintained by a water ejector pump, was 10 mm. After this evaporation process the filter shell was weighed and the brittle cake removed by a spatula. The filter was then suspended in an electric furnace and the remaining carbon burned. The difference in weight between the loaded and the clean filter was used as a basis for the calculation of the amount of suspended phase in the tar. However, in most cases a correction had to be made for the volatile material still present in the cake. This correction factor was determined by placing a known quantity of the detached cake in a beaker and heating on a hot plate to 250-300°C. for 2 4 hours, after which there was no further visible evolution of vapors from the cake, The amount of volatile matter lost during this treatment was "determined and used as correction factor. The amount of such vapor loss was usually less than 5 per cent of the weight of the cake. Duplicate or triplicate tests were run on each sample. With seven tars, checks of 5 per cent were obtained. In the case of the two high-carbon tars the checks were within 10 per cent. The results are presented in the following table:

water and to bring their consistencies to a common value, one gallon (3.8 liters) of each tar was distilled by direct heating. in a 5-liter balloon flask, until the residue had a viscosity corresponding to a float test (A. S. T. M. method D 139-27) of about 180 seconds a t 32" C. The value of 180 seconds for the standard float test was selected because it represented the consistency of the dehydrated sample 9-H, the most viscous of the lot. I n the range of commercially useful tars this consistency is characteristic of one of the more fluid grades. The following table gives the amount of distillate removed to bring each tar to the indicated consistency: * Sample No.

Viscosity by Float Test at 32' C. Seconds 186.5 182.9 183.1 182.9 189.5 184.6 187.5 184.0 180.0

Type .. of Tar

1-v 2-V

Vertical-retort

8-H 9-H

Horizontal-re tort

723

Distillate Removed Per cent b y w t . 15.9 15.5 12.0 4.0 2.3 15.0 5.1 4.1 0.0

Sample No

To study the effect which the suspended solid material, observable under the microscope, has on physical characteristics, approximately one liter of each distilled tar was filtered. A specially constructed pressure filter submerged in a hot water bath was used (Figure 1). The filter medium, E, consisted of an alundum extraction shell of medium porosity (Norton R H 360-34 mm. diameter, 100 mm. high), which was cemented into the brass collar, C, by means of a paste of zinc oxide in water glass. The collar fitted snugly between the upper and the lower part of the outer cylinder, 0. Lead gaskets were placed above and underneath the collar. After placing heated tar in the filter, the cap, P , was screwed into place and compressed sir was admitted through the connection, L. The operating pressure was usually kept close to 50 pounds per square inch (3.5 kg. per sq. cm.). The filtrate was collected in a 500-cc. Erlenmeyer flask immersed in an ice'bath. In spite of the relatively large filter area and the pressure employed, the operation of the filter was exceedingly slow. On the average, 150 t o 200 cc. could be filtered before the cake had to be removed and the filter cleaned. The cycle required 6 to 8 hours. After the end of the filtration, the cake was treated with two successive washings of 75 cc. each of nitrobenzene, and finally with 100 cc. of benzene. This also was done under pressure

Per Cent by Wt. Removed by Filtration

Type of Tar

1-v

Vertical-retort

3-c

Coke-oven

2-v

4-c 5-c

3.6 3.4 1.0 1.3

4.0 4.2 5.7 23.9 28.9

6-C

7-c 8-H

Horizontal-retort

9-H

Analysis of Unfiltered and Filtered Tar Samples All experimental work in this investigation was carried out on the eighteen samples (nine unfiltered and nine filtered distilled tars) prepared in the manner described. I n order to determine their main characteristics and to make possible their grouping within the various classes of industrial products, the samples were analyzed with respect to specific gravity (A. S. T. M. method D 70-27), float test a t 32" C. (A. S. T. M. D 139-27), distillation (A. S. T. M. D 20-30), softening point of the distillation residue (A. S. T. M.

AND OPERATINQ CONDITIONS FOR PRODUCTION OF TARS TABLE I. TYPEOF DQTJIPMENT

Sample Type of Cprbonization No. EauiDment

No. and Type of Ovens or Retorts

Capacit of Individlal Av. Oven Width or Retort of Oven T o n s per charge Inches

Taper Inches

Height Feet: inches

Length Feet :inches

P F

1-v 2-v

Continuous retort Continuous retort

vertical

16 Glover West 32 Glover West

6 . 5 per day 3 . 5 per day

vertical

16 Glover West

6 . 5 per day

.. ..

Top: 39'/2-X 91/z in: Bottom: 40' z X 24 In. Length: 25 Top: 33 X l o i n . Bottom: 39 X 18in. Length: 25 ft.

/t.

. . ... .....

A t top: 1300 At discharge: 2400

1

11.15 16.9

16 16

2 2

9 : 1Ob/s 13:6

37:O 47:21/2

57 Becker 42 Becker

15.4 6.81 11.75

14 131,'t 167/8 17

2

13:6 1O:lO

42 : 21/2 23 :81/2 30:10'/2 36:O

By-product ooke oven

43 Becker 61 Becker 100 Koppers

12.6 12.5 13.25

8-H

Horiaontal retort

100

9-H

Horizontal retort

By-product coke oven

4-c 5-c 6-C

By-product coke oven By-product coke oven By-product coke oven

7-C

'

100 Koppers 80 Semet Solvay

10 benches (Parker Russell M. and

M. Co.)

11 benches (Russell Engineering Go.)

.. Av. charge, 1700 Ib. per retort

..

8/4

11/2

2

11:81/z 9 :5/8

Temp. Conditione ai Reported by Plants F. Av. in flues, 2450

Koppers 1100 49 Becker

3-c

Coking Time

Airport: 2050 Walls: 1860 Horizontal flues: 1700 No record Av. carbonizing temp., 2260 No record Flue: 2300 Temp. ooke: 1800 90% of coke product in 14-in ovens

Modified D shape, 8 retorts per bench Height: 16 in. Width: 26 in. Len th: 21 f t . 9 in. D s l a p e , 8 retorts per bench Height: 16 in. Width: 26 in. Length: 22 ft.

.... .

No record

. ..,.

Setting: 2000

INDUSTRIAL AND ENGINEERING CHEMISTRY

724

VOL. 28, NO. 6

TABLE11. ANALYTICAL DATA Sample No. Sp. gr. a t 25'/25' C. Float test at 32' C., sec. Distn., per cent by wt. to: 170' C. 2100

c.

235' C. 270' C. 300' C. Residue Softening point5 of residue, Total bitumen

Sp. gr. at 25'/25' C. Float test at 32" C., sec. Distn. per cent by wt. to: 170d C . 210" c. 2350 c . 270' C . 300: C Residue Softening pointa of residue, Total bitumen a Ring and ball.

C.

O

c.

1-v

2-v

1,154 166,5

1.166 162.4

0.0

0.0 0.2 4.7 13.7 86.2 43,5 8.43

0.0 0.0 0.2 6.2 14.9 85.0 45.7 6.56

1.145 160.7

1,150 168.7

0 0 0.0 0.2 4.2 13.7 86.2 40.1 5 61

0.0 0.0 0.0 6.3 16,s 82.9 44.4 3,l2

3-C 4-c Unfiltered Tars 1.175 1.195 184,6 181.9

.

5-c

6-C

7-c

8-H

1.207 182.1

1.215 173.9

1.218 183.7

1.241 184.3

1.270 176.3

0.0 0.0

0.0 0.1

9-H

0.0 0.0 0.0 1.5 1.7 10.0 8.5 16.1 15.5 83.6 84.3 49.0 55.2 4.33 7.09 Filtered Tars 1.172 1.193 188.0 187.2

0.0 0.1 1.5 7.8 13.4 86.5 52.0 8.03

0.0 3.9 10.2 89.6 39.6 7.44

0.0 0.0 1.9 8.6 13.6 86.3 51.4 11.17

0.0 0.0 1.2 7.3 14.4 85.5 54.5 23.02

12.2 18.9 80.9 07.6 28.96

1.202 171.2

1,209 167.8

1.2075 186.2

1.196 134.6

1.202 110.1

0.0 0.0 1.1 8.2 16.4 83.5 53.3 5.56

0.0 0.0 1.8 8.4 17.0 82.9 52.2 6.24

0.0 0.0 0.0 3.4 11.3 88.6 40.1 4.21

0.0 0.0 2.6 10.1 17.3 82.7 56.7 6.79

0.0 0.0 2.2 10.5 18.9 81.0 53.7 6.52

0.0 0.0 7.7 15.6 22.9 76.8 59.4 8.53

0 0

0.0 0.0

0.3 6.9 17.1 82.4 49.0 3 36

D 36-26), and total bitumen (A. S. T. M. D 4-27). The results of this analytical work, presented in Takle 11, are self-explanatory. The float-test values of the unfiltered tars do not coincide exactly with those given in connection with the preparation of the tars. In all instances where this difference is greater than the experimental error, it may be due to changes which had taken place within the tars during the 4-month period between the preparation and the testing of the samples. Contrary to expectations some float-test values of the filtered tars are slightly higher than thosp of the corresponding unfiltered tars. This can be attributed to evaporation losses during the filtration, as may be seen from the distillation data.

Suspended Material A quantitative microscopic study was undertaken to determine the number of particles suspended in a unit weight of tar and also their size distribution. A few days after the preparation of the tars, 0.1-gram samples were dispersed in 50 cc. of nitrobenzene. Microscopic analyses of these samples showed that this concentration was a practical one for the purpose of counting. However, a concentration of 0.1 gram in 100 cc. of nitrobenzene was found more suitable in the case of the horizontal-retort tars. These procedures do not coincide with those recommended by Nellensteijn (9). They were chosen primarily for the convenience of counting and sizing. By careful checks it was established that the concentration affected the results of counts only in so far as high concentration made it difficult to distinguish, within a given field, between particles already accounted for and those still to be counted. The occurrence of agglomerates in dispersiocs of tar in nitrobenzene is a

5.0

serious handicap to a reliable examination of the dispersed phase. To overcome this difficulty, a procedure was employed which permits a close duplication of counts made on dispersions from the same sample. This procedure calls for the violent shaking of the dispersion a t room temperature over a period of a t least 1.5 minutes. To assure sufficient agitation, an Erlenmeyer flask having twice the volume of the dispersion is used. This method was found satisfactory. The counts were made immediately after dispersion. A few drops were brought onto a Thoma hemocytometer which was then placed on the stage of the microscope. The magnifying power of the optical system used for counting was 400 diameters. After a period of 30 minutes, within which most of the particles settled t o the bottom of the chamber, the microscopic image was projected onto a screen and the number of particles counted within fields of known area. Special care was taken in counting not only the particles adhering to the bottom of the chamber, but also those which were still in Brownian movement and those which adhered to the cover slide. An average of twenty-five fields was counted in each case. Duplicate samples checked within less than 10 per cent, The number of particles in each sample of unfiltered tar is expressed as number per milligram of tar and is as follows; the values may be converted to the units used by Nellensteijn (9) by multiplying by the respective specific gravities of the tars: Sample No. 1-v 2-V 3-c 4-c 5-c 6-C 7-c

8-Is 9-Is

Type of Tar Vertical-retort

Coke-oven

Horizontal-retort

No. of Particles per Mg. of Tar 39 700 000 53:500:000 19;200;000 42,800,000 75,200,000 114,000,000 91,200.000 281,000,000 107,000,000

The determination of particle size and size distribution was made by measuring the size of 500 to 1000 particles in each dispersion. Again the projecting microscope was employed, using a magnification of 625 diameters. The projected image on the screen was approximately 5000 diameters. The results of the size measurements are illustrated by Figures 2 and 3. In Figure 2 cumulative per cent by weight of undersize is plotted against particle diameter. To show more clearly the differences in the size distribution of the suspended phase, a more sensitive form of representation is given in Figure 3,

INDUSTRIAL AND ENGINEERING CHEMISTRY

JUNE, 1936

a-b c-1 = b

where per cent by weight for 0.2-micron intervals is plotted as ordinate and particle diameter as abscissa (21). The determination of size distribution made it possible to calculate the total weight of the solid material suspended in the individual tars (6). The value,

a = (C-I

b =

s

Suspended Phase Per Cent by Wt. A1 Unfiltered Tar 5.6 3.9 0.6 1.4 3.8 4.0 5.6 24.4 28.2

Type of Tar Vertioal-retort Coke-oven

Horizontal-retort

The main results of the microscopic investigation are summarized in Table 111 in which the various tars are arranged according to the fineness of their solid phase. The table also shows the amount of suspended phase present in the unfiltered tars as calculated from the total bitumen of each unfiltered and filtered tar. This was done in a manner similar to that outlined by Hodurek (7). The following formula was used : ~

-

~

‘-I1 100

+ C-11)

l-m

X 100 =

% ’ insol. in unfiltered tar

loo - % insol. in filtered t,ar - ((3-1) -

where C-I = suspended phase C-I1 = solvent-insol. - (C-I)

where n is the number of particles wibhin a given interval and d, their diameter, represents the total volume of all particles measured, provided their shape is spherical. This value was obtained by adding for each tar the apparent volumes of each interval and multiplying the sum by 7r/6. By relating the n u m b e r of particles sized to the total number of particles present in a milligram of tar, the total v o l u m e of t h e suspended phase was computed. F i n a l l y , b y multiplying this total volume by the specific g r a v i t y of t h e suspended phase, t h e weight per cent of the suspended phase in the tar was obtained. A specific gravity of 1.5 was used, a value which was found to be the average of the suspended phase as calculated from the difference of the gravities of the unfiltered and filtered tars. According to this procedure the following concentration of the suspended phase, expressed in weight per cent of the unfiltered tar, was calculated for the individual samples : Sample No. 1-v 2-v 3-c 4-c 5-c 0-c 7-c 8-H 9-H

725

~~

~~

~

The symbols “C-I” for the suspended phase, and “C-11” for the solvent-insoluble material exclusive of the suspended phase, were proposed by Hodurek ( 7 ) . They will also be used in this paper. Referring to Table 111, the C-I values determined by filtration show good agreement with those determined by Hodurek’s method in the case of the tars low in C-I. When C-I is present in large amounts, the results are consistently high. This is due to the difficulty in the purification and drying of the filter cake. The results of the microscopic analyses are also high. This is to be expected since,. in the computation of the per cent C-I as an approximation all particles are assumed to be spheres. Furthermore, average diameters are used which represent the arithmetic instead of the cubical mean. An inspection of the results of the microscopic mea~urements confirms the known fact that the concentration of the suspended phase is considerably higher in horizontal-retort tars than in vertical-retort or coke-oven tars. The number of suspended particles in the horizontal-retort tars is also greater than that of the other two groups. However, the three types differ in the size distribution of their suspended phase. The coke-oven tars have a narrow particle-size range. Next come the continuous vertical-retort tars, the range of which extends to an upper limit of 4 to 5 microns. Finally, the frequency of coarse particles is most pronounced in the case of the horizontal-retort tars. In the case of vertical-retort tars derived from intermittent operations, a particle-size range even smaller than that of coke-oven tars may reasonably be expected.

Solvent-Insoluble Materials The tests so far reported were performed to determine the amount of suspended material in each of the tars. In this section experiments are described which were conducted to determine the concentrations of those constituents which precipitate when tars are brought into contact with certain solvents. Since a direct comparison of the action of various solvents on the tars is of interest, an investigation was undertaken to determine by the same method and a t the same temperature the amount of solvent-insoluble material (C-11) present in each of them. Benzene, acetone, ethyl ether, and petroleum ether (40-60” C., Eimer and Amend) were used as solvents. To eliminate the interference of the suspended phase, all tests were carried out on the filtered tars.

~~~~~~~

~

OF MICROSCOPIC INVESTIQATION TABLE 111. SUMMARY Particle Diam. in -Microns at Following Per Cent by Wt.: 25 50 75 100 0.8 1.0 . 1.2 1.8 1.3 2.0 0.7 1.0 0.9 1.1 1.4 2.4 0.8 1.1 1.4 4.0 1.8 3.0 0.9 1.3

Sample No. 4-c 3-c 5-c 0-c 7-c

S o . Particles per Mg. Tar 42,800,000 19,200,000 75,200,000 114,000,000 91,200,000

Vertical-retort

2-V 1-v

53,500,000 39,700,000

1.0 1.3

1.4 1.8

1.9 2.2

5.0 4.0

Coarse Coarse

Horizontal-retort

9-H 8-H

281,000,000 107,000,000

1.6

1.4

1.9 2.4

2.4 3.5

4.0 0.0

Very coarse Very coarse

Type of Tar Coke-oven

Classification Very fine Very fiiie Very fine Fine Fine

Suspended Material Per Cent by Wt. of Unfi1te;ed Tar BY BY By Hoduiek’s filtration microscope method 1.3 1.4 1.62 1.0 0.5 1.00 4.0 3.8 1.91 4.2 4.0 3.37 5.7 5.6 4.71 3.4 3.6

3.9 5.5

3.66 2.99

28.9 23.9

28.2 24.4

22.32 17.66

726

INDUSTRIAL AND ENGINEERING CHEMISTRY

First it was necessary to find a method that would permit the use of all solvents a t the same temperature conditions. To investigate the suitability of accepted extraction procedures, 0.7 gram of a filtered tar was placed in an alundum extraction shell (Norton RA. 360) of known weight. The thimble was then suspended in a Soxhlet extraction apparatus, and the tar extracted with 150 cc. of c. P. acetone. Heat was supplied by an electric hot plate. After 9, 21, and 28 hours of total extraction time, the shell was removed, dried in a n oven (110' C.) for 30 minutes, and weighed. The following results were obtained: after 9 hours, 26.0 per cent residue: after 21 hours, 16.9 per cent; and after 28 hours, 14.4 per cent.

At the end of the 28-hour extraction period the filtrate (leaving the shell) still showed a distinct yellow discoloration. This indicated that the extraction had not been completed. The slowness of the extraction method as well as the uncertainty of a definite end point made this method impractical for adoption in this investigation. Next, the A. S. T. M. method for the determination of bitumen was tried and again acetone was used instead of carbon disulfide as specified in that test procedure. Contrary to the rapid solution in carbon disulfide, the tar did not dissolve readily. Stirring with a rod was finally tried. This resulted in breaking up the original deposit. The individual droplets of the tar, however, adhered tenaciously to the walls of the flask and to the stirring rod. After filtration, flask and rod were also dried in a n oven and the amount of residue adhering to the glass was determined. However, results could not be duplicated. An examination of the residue showed that in the case of the larger droplets a hard skin had been formed on the outside. Underneath this hard layer, unattacked tar, recognized by its bright luster, could often be observed. On account of the failure of the two standard methods to satisfy the conditions required by the work undertaken, a new method was developed which has the following advantages: (1) It furnishes reproducible results, and (2) it provides a well-defined basis for the comparison of the specific solvent power of various solvents. In principle, the method consists of first producing a well-distributed thin film of tar which is then digested in a given volume of solvent. The determination of the insoluble residue is finally performed in much the same way as outlined in A. S. T. M. procedure D 4-27: The thin film of tar is produced on a circular section of finemesh wire gauze (2.5 cm. in diameter) t o the center of which is attached a copper wire. This copper wire serves t o suspend tbe tar specimen in the solvent. Figure 4 illustrates the construction of this tar carrier. After the carrier and the crucible through which the final solution is to be filtered have been weighed, a sample of the tar

VOL. 28, NO. 6

under investigation is heated in a beaker approximately 3 om. in diameter to a temperature at which the tar is moderately liquid. The carrier is then lowered into the tar until the wire gauze is covered completely by it. It is then raised out of the tar, and the excess tar adhering to the gauze is removed by rapidly twirling the wire between thumb and forefinger. If this is done properly, the coating of tar left on the gauze is so thin that it is possible to distinguish the form of the individual wires of the wire gauze. To determine the weight of tar adhering to the gauze, carrier and Gooch crucible are weighed again. From the actual weight of the tar on the carrier, the amount of solvent t o be used is calculated. The dimensions of the equipment used and a series of tests showed the ratio of 0.5 gram of tar to 200 cc. of solvent to be practical. The amount of solvent required by this ratio is finally transferred into a tall 500-cc. beaker. The carrier is attached to a hook provided on the cork stopper of the beaker, and the stopper brought into position. An air-driven stirrer, also held in position by the cork stopper, is started and the digestion continued for 45 to 60 minutes. Figure 4 gives a view of the assembled digestion apparatus. After digestion has been completed, the carrier is placed in the Gooch crucible and the solution is filtered. If small particles adhere t o the walls of the tall beaker, the filtrate is transferred back into the digestion flask, agitated, and poured again through the filter. If neceswary this procedure is repeated. Further purification of the solid residue in the crucible by washing with pure solvents was not thought advisable since they may have solvent powers different from those solvents in which some tar constituents have been dissolved. This holds true particularly in the case of solvents in which tars are soluble to only a relatively small extent. Consequently, after completion of filtration, crucible and carrier were placed into a drying oven (110' C.) where they were kept over a period of 6 to 10 minutes. When petroleum ether was used as a solvent, the drying period could not be extended over more than 5 minutes, since the residue still contained so much volatile material that prolonged heating caused evaporation of its more volatile portions. Before adopting this new method for the quantitative analysis of the resinous constituents in the tar samples, four series of tests were performed to determine: (1) the reproducibility of the test, (2) the relation between the amount determined to be insoluble in a given solvent and the temperature of the solvent during the period of digestion, (3) the effect of changing the ratio between the amount of tar and the volume of solvent on the percentage of insoluble residue, and (4)the effect of the time of extraction on the amount of insoluble residue. The reproducibility can be demonstrated by the following two sets of typical results: Sample

No.

0-C

No: of

Determinations 2 3

Residue 19.74 19.96 19.44 19.70 19.71

1

27.22 27.43

1

4 Mean 9-H

Per Cent

2 3 4 Mean

26.75 20.90

27.07

Deviation from Mean 0.03 0.25 0.27

0.01 0.14

0.15 0.30 0.32

0.17 0.26

The mean deviation in the first case is 0.14 or 0.7 per cent, in the second case 0.25 or 0.9 per cent. The average deviation is =t0.35 per cent in the first and * 0.45 per cent in the second case. The maximum deviation is 1.4 per cent in the first and 1.3 per cent in the second case. If the limitations of the test are taken into account, such as lack of washing of the solid residue, these duplicate results can be considered satisfactory. Besides, an examination of the filtered solutions showed a marked Tyndall effect. Microscopic examination of the filtrate showed a few visible particles of a size between 0.2 and 1 micron. This indicates that some of the dissolved material is present in the form of a colloidal suspension. The degree of dispersion of this suspension and the conditions of the filter mat should also have a n influence on the precision of the test. It is of interest to note that the

JUNE, 1936

INDUSTRIAL AND EN(XNEERING CHEMISTRY

precision of the test increases with the solvent power of the solvent and decreases with increase of the amount of solid residue. The influence of temperature on the amount of insoluble residue was studied on sample 9-H by performing extraction tests with acetone for 60 minutes a t three different temperature ranges. The results of these tests are as follows:

727

between the amount dissolved in acetone and the insoluble residue was studied on two tars. The results of the tests are as follows: Sample No.

Extn. Time Min.

Insol. Residue

Sample No.

%

Extn. Time .Win.

Insol. Residue

%

Av. Acetone-Insol. Av. Scetone-Insol. Av. Acetone-Insol, Temp. of Residue Temp. of Residue Temp. of Residue Solvent Remaining Solvent Remaining Solvent Remaining

c. 0 +2

%

a

24 71 24 82

c. 26 26

c.

% 26.90 26.75

47 49

% 25 99 25 69

The amount of insoluble residue decreases with increasing as well as decreasing temperatures. The reason for this peculiar behavior can perhaps be found in the state of dispersion of the insoluble material in the solvent. While a t higher temperature (47’ to 49” C.) the solvent power of the acetone is actually increased, resulting in a lower residue than a t room temperature, the extraction a t lower temperature yields also a decreased residue because the solid particles, dispersed in the solvent, are of smaller size. The phenomenon is comparable to the growth of crystals in hot solutions. It d a y therefore be said that the effect of normal temperature variations on the amount of insoluble residue is small. At ordinary room temperatures precautions to maintain a constant solvent temperature are not warranted. A number of tests were performed to determine the effect of changing the ratio of the quantity of tar to the volume of solvent. Since the apparatus as described does not allow great variation in the amount of tar used, additional tar carriers of smaller size were constructed. The smallest one had a diameter of approximately 1.5 cm. The results of the tests are given in the following table; to facilitate comparison, the various ratios were reduced to milligrams of tar in 100 cc. of acetone : Per Cent by Wt. Insol. Residue 29.97 29.72 29.25 27.70

Mg.

Test

No.

Tar er 100 8 c . Acetone 500 249 145 80

Mg.

Test No. 5 6 7

Tar er 100 Aoetone 77 40 18 14

8,.

Per Cent by Wt. Inspl. Residue 27.50 25.60 25.93 25.26

It is evident that 45 to 60 minutes are required to obtain an equilibrium between the dissolved and the solid phase, provided a moderate amount of agitation is employed. ‘uh

51

$G

2:

bQ

$$

%* 5% All filtered tars were analyzed according to the procedure outlined. The results for the several solvents are summarized in Table IV. The figures given for the unfiltered tars represent the amount of C-I1 in per cent by weight of the unfiltered tars. The conversion froin the percentage insoluble in the filtered tars to that of the unfiltered tars was made by means of the formula :

C-I1

(

Z)

R X 1 -where C-I1 = per cent insoluble in unfiltered tar, exclusive of suspended phase R = per cent insoluble as determined on filtered tam C-I = per cent suspended solid material as determined by Hodurek’s method with CS2as solvent

For coinparison Table IV also contains the results of the percentage insoluble in carbon disulfide as determined by A. S. T. &.I.procedure D 4-27. The new method differs from 8 this procedure only in that a thin coating of tar is exgosed to the action of the solvent instead of a-droplet. A airect The results are plotted in Figure 5. Thus concentrations comparison is therefore justified when a solvent is used above 200 mg. per 100 cc. of acetone give practically the same which dissolves the tar readily. values. The results show that, in general, the ainouiit insoluble The length of time required to arrive a t a n equilibrium in a solvent increases as the surface tension of the solvent decreases. The molecular structure of the solvent also has T A 4 R L E I\’. RESULTSO F SOLUBILITY DETERMINATTONS an i m p o r t a n t b e a r i n g as indicated by the --Per Cent by Weight Insoluble in.---Sample Condition Carbon Ethyl petroleum amounts of c-11 insoluble in ethyl ether and Type of ‘ h i No. of Tal disulfide Benzene Acetone ethei ether in petroleum ether. Both solvents have pracVertical-retort 1-v 68.4 Unfiltered 5,44 7.43 s 97 11.09 tically the same surface tension. However, Filtered 5.61 70.5 7.66 9.25 11.43 lacking knowledge of the exact composition of 72.5 2-v Unfiltered 3.01 4.93 5.99 9.04 tars, it is not possible to use molecular structure 75.2 Filtered 3,12 5.11 6.21 9.37 as a basis for predicting accurately the solubili78.2 Coke-oven 3-c Unfiltered 3,32 4.88 7.34 12.17 Filtered 3.36 79.1 4.93 7.42 12.30 ties of tars in various solvents. For practical purposes the surface tension values of solvents 4-C Unfiltered 5.47 80.4 9.46 15.66 26.1 81.7 Filtered 5.56 15.92 9.62 26.5 give a good indication of solubility characteris5-c 80.6 Unfiltered 6.12 IO.66 21.1 tics. 26.5 1 2 3 4

Filtered

6.24

10.87

Unfiltered Filtered

4.07 4.21

8.99 9.30

7-c

Unfiltered Filtered

6.47 6.79

11.41 11.97

8-H

Unfiltered Filtered

5.37 6.52

7.92 9.62

9-H

Unfiltered Filtered

6.63 8.53

11.93 15.36

6-C

Horizontal-retort

21.5

27.0

82.2

19.02 19.69

26.2 27.1

80.0

25.2 26.4

31.2 32.7

77.5 81.3

11.62 14.12

18.5 22.5

64.5 78.4

20.8 26.8

23.1 29.7

63.2 81.3

82.8

Viscosity Index For evaluating viscosity indices it was necessary to determine the viscosity of each sample over a wide range of temperature. Obviously the standard efflux type instrument could not be used for this purpose. Tests on a MacMichael viscometer showed that its results were reliable only for comparatively fluid tars. I t was used in the

INDUSTRIAL AND ENGINEERING CHEMISTRY

728

range above 60' C , For viscosity determinations below this temperature the capillary rise method as described by Pochettino was employed (14). The apparatus used for this purpose is shown in Figure 6. A few cubic centimeters of tar were placed into the test tube (2.5 cm. in diameter), the test tube was inserted into the constant-temperature flask, and the capillary tube was brought into position. Thirty minutes were allowed for the equalization of temperature. I n the meantime, while stopcock 2 was kept closed, the reservoir wm evacuated to a fixed value of absolute pressure in the range 600 to 300 mm. of mercury. Finally, after stopcock 1 had been shut, stopcock 2 was opened and the progress of the rising tar-air interface in the capillary recorded with the aid of a stop watch. From the rise of the tar in the capillary, Pochettino (14) calculated the true viscosity by means of the formula: 9 =

E

where I ! r = At = =

ZI =

g X

4

(122

9 =

n X g X r 4 X [H-p(E-h)] 8 X V X Z

- where

H p

X t

pressure as indicated on pressure gage, grams/ sq. om. = sp. gr. of liquid =

I n setting up the differential equation for small increments of 1 and t and neglecting increments of the second order, the following equation is obtained :

- 112)

The formula is a direct derivation of Poiseuille's equation for the flow through capillary tubes. I n substituting for driving force H , Pochettino used the pressure as observed on the pressure gage. For relatively high pressures this is undoubtedly sufficiently accurate. However, if relatively fluid materials are to be tested, which in some instances was the case in the present investigation, low pressures are more convenient because, for a given distance of rise, they result in longer time intervals. A correction was applied to the factor H , as read on the pressure gage, to account for the static head of the tar in the capillary. For extremely low driving forces the pressure exerted by capillary forces on the tar in the tube should also be taken into consideration. However, in no case was this last correction warranted. The fdrmula to account for the change in static head in the capillary was derived in the following manner: A

of the atmosphere, H , from which is subtracted the pressure exerted by the static head of the fluid. Therefore, with reference to the symbols indicated on the accompanying sketch the formula should be written:

H X r2 X At

x

acceleration due to gravity driving force radius of capillary time interval, seconds length of capillary covered by tar at end of period length of capillary covered by tar at beginning of period

'=

VOL. 28, NO. 6

X g X p X r4 X 1 8 X V X 1 .

and represents Poiseuille's equation for the flow of liquids through capillary tubes. In the case of a liquid rising through a capillary tube, the driving force, p , in pressure units, equals the difference between the pressure in the capillary and that

I

Since V = T X r2 x 1, the equation can be simplified to:

Integrating between the limits 11 and

or 11 - 12

+PA + HP

H IgH

k:

++ PAph -- 11 = g X r2 X

P X (ta 8XI)

12

- k)

Simplified and transformed into an expression for 7, the desired formula is obtained: g

x

T2

x

p

x

(k

- a)

Before viscosities could be evaluated from the formula, the radii of the capillary tubes used for the experiments had to be determined. This was done by filling the tubes to two definite levels with mercury and subsequently weighing the amount of mercury contained in the tubes. The diameter determined in this way for each tube is the average of two determinations. I n each case checks of less than one part in a thousand were obtained. The diameters were found to be as follows: capillary 1, 0.0908 cm.; capillary 2, 0.0916; capillary 3, 0,0901. The capillary tube method was used to determine the viscosity of the tars a t about 30" and 40" C . The following results were obtained: Sample NO.

Vertical-retort

1-v

29.1 39.8

842 131

26.7 37.5

1547 191

2-v

28.9 40.1

958 127

27.0 37.5

1530 204

3-c

28.4 40.1

1780 184

27.5 37.8

2180 280

4-c

28.6 40.2 50.2 65.6

1660 185 40.4 7.16

27.0 37.4

2770 333

5-c

29.4 40.4

1490 183

27.4 37.5

2790 376

6-C

28.9 39.8

1353 160.2

26.7 37.5

2225 245

7-c

28.8 40.1

1815 207

26.4 37.5

2880 328

8-H

28.3 40.1

1960 213

27.5 37.1

955 153

9-H

29.5 39.8

1200 246

27.5 37.0

566 118

By-product coke-oven

Horisontalretort

Unfiltered Tar Temp. Viscosity

Filtered Tar Temp. Viscosity

Type of Tar

c.

Poises

c.

.. ..

Poises

.. ..

JUNE, 1936

INDUSTRlAL AND ENGINEERING CHEMISTRY

The MacMichael determinations were performed on a standard instrument equipped with a wire of gage No. 20. The instrument was standardized by means of a glycerol solution, the viscosity of which had previously been determined by the capillary rise method. A deflection of the wire of one scale division was found to be equivalent to 0.90 poise. This calibration was further checked by determining the viscosity of tar 4 at 65.5' C. by means of the capillary rise method. The resulting value of 7.2 poises coincides within 10 per cent with the value of 7.84 determined with the MacMichael viscometer. This check is to be considered satisfactory since, owing to the short interval of rise (13 seconds) at this viscosity, the capillary rise method is estimated to be accurate to not more than 10 per cent. The MacMichael viscometer tests of the various tars were performed by first preheating the tar t o approximately 80' C. A volume of 100 cc. was then transferred to the cup of the viscometer which had been preheated by hot water in the heating jacket. The motor was started. and a period of 5 to 10 minutes allowed for equalization of temperature conditions. After this period the motor was temporarily stopped and the temperature of the tar ascertained. Immediately after removal of the thermometer and start of the motor, a deflection reading W R S taken. After another period of 5 t o 10 minutes within which the temperature of the tar was allowed to drop a few degrees, another set of readings was taken and this procedure repeated until a temperature of approximately 60" C. was reached. Readings taken below this temperature were usually unreliable, since temperature variations could be observed between varying points of the cup. Consequently, these readings were discarded. The results of the MacMichael viscosity determinations are as follows: Type of Tar Vertical-retort

By-product cokeoven

Horizontal-retort

Sample NO. 1-v

Unfiltered Tars Temp. Viscosity c. Poises 72.0 4.7 9.2 63.6

Filtered Tars Temp. Viscosity O C. Poises 65.6 7.5 63.5 5.4 62.5 10.3

..

..

2-V

64.0 59.2

9.8 15.1

65.9 64.1

5.2 8.1

3-c

66.7 65.1 62.6 61.3

6.5 7.6 11.0 12.1

.. ..

..

..

4-c

65.6 62.5 60.3

7.8 11.1 16.6

66.6 63.9 60.2

7.b 11.0 16.4

5-C

63.8 61.1

11.1 13.3

..

69.0 65.1 61.2

7.6 11.6 18.4

6-C

65.6 63.6

7.2 7.5

61.5 59.7

9.2 12.1

7-C

67.3 66.4 64.4 60.6 60.0

8.0 9.7 11.7 17.1 18.0

68.6 64.0 60.6

7.4 11.6 17.7

8-H

67.3 66.1 63.1

8.8 11.5 17.8

66.0 62.5 58.6

4.8 7.4 11.4

9-H

70.3 66.7 62.9 59.3

11.5 16.9 23.2 32.8

67.2 62.5 58.9 54.9

4.4 7.7 11.1 16.0

..

.

I

.. ..

.. ..

..

.. ..

Evans and Pickard (2) and later Klinkmann (8) and Eymann (3) pointed out that, in the case of most bituminous materials, the relation between viscosity and temperature can be expressed by a simple exponential function of the form q x t" = a, where q and t are viscosity and temperature, and n and a represent two constants. If this relationship is correct, a straight-line function must result if the logarithm of viscosity is plotted against the logarithm of temperature. Provided the logarithm of temperature is plotted as ordinate and the logarithm of viscosity as abscissa, then the tangent of the angle which the straight link forms with the abscissa is a measure of the change of viscosity with temperature. Evans and Pickard expressed temperature in Fahrenheit

729

units, whereas Klinkmann and Eymann preferred to use the centigrade scale. This makes a difference in the absolute values of the index. The results of the present investigation indicate that American coal tars follow the same viscosity relationship which has been established for European tars. I n the case of each tar, regardless of whether unfiltered or filtered, a straight line is obtained when the logarithm of temperature (centigrade scale) is plotted against the logarithm of viscosity. In general, the viscosities of the filtered tars are of the same magnitude as those of the corresponding unfiltered tars. Only the horizontal-retort tars differ in this respect. Their filtered tars are more fluid. Since the viscosity tests were performed mainly to study the changes of viscosity with temperature, it was of interest to determine whether the viscosity characteristics of the filtered horizontal retort tars would change if they were distilled to the absolute viscosity range of the other tars. A small sample of filtered tar (9-H) was distilled to a float test a t 32" C. of 215 seconds. The resulting residue was then tested by the capillary rise method which gave the following results: Temp., O C. 29.2 37.4 44 2

Viscosity, Poise9 1740 377 128

The susceptibility of viscosity to temperature changes is substantially the same for each group of the tested tars. The observed variations may be noted from the following table which shows the values

- log 72 log t z - log t l

log 71

for each tar, where q is given in poises and t in C . ; no distinction is made between the unfiltered and filtered tars since both have the same slope. Sample No. 1-V 2-V 3-c 4-c 5-c

Value 5.94 6 05 6 53 6 55 6 49

Sample No 6-C 7-c 8-H 9-H 9-H (filtered) distilled

Value 6 50 6 3h 6 03 5 26 6 30

The temperature coefficient of viscosity is highest in the case of the coke-oven tars and lowest in the horizontal-retort tars. However, the smaller change of viscosity with temperature in the horizontal-retort tar is due to the fluidity of their liquid phase, as may be seen from the values for tar 9-H. By distilling the filtered sample of this tar to the consistency range of the unfiltered tars, the coefficient was increased from 5.26 to 6.30. From the standpoint of the present investigation i t is of immediate interest to note that: 1. The solid material suspended in coal tars may have a stiffening effect upon the liquid phase. In the case of each tar, the lines representing the filtered and unfiltered tar have exactly the same slope within the limits of experimental error. 2. The stiffening effect of the suspended phase becomes appreciable only in the case of the horizontal-retort tars in which the solid phase has a concentration of approximately 20 per cent.

A further discussion of the significance of the viscosity determinations will be taken u p later.

Surface Tension As mentioned previously, determinations on coal tars were carried out by Nellensteijn (II), Klinkmann (8),and Fricke and Meyring (4). These investigators used the following methods : maximum bubble pressure, capillary height, force of adhesion, and stalagmometer. Since the capillary height method presupposes a zero contact angle,

INDUSTRIAL AND ENGINEERING CHEMISTRY

730

a condition which is not fulfilled in the case of tars (4), this method was eliminated. Klinkmann's method of determining the force of adhesion and the stalagmometer are both restricted to relatively limited temperature ranges. For these reasons, the maximum bubble pressure appeared to be the most suitable method, particularly since Nellensteijn had used it on tars within a temperature range of 120" to 40" C.

AL'MVONsN//P BfTWE€N SUMAC€ 7€NJ/ON AND 7€MP€RAZW€ D€T€RMN€D ON UUi,Q€D TAR (m BY BUBBLLC fXESUR€ AND DU NOW WNS/&?fCCR

The use of this method on the tars under consideration gave easily reproducible results a t the maximum temperatures employed between 80" and 90" C. At lower temperatures inconsistencies appeared which destroyed the continuity of the results obtained a t higher temperatures. Although Nellensteijn's results show also a sudden break in the change of surface tension with temperature, it was decided to check the surface tension of the tars a t lower temperature by means of an instrument capable of eliminating the possible influence of the change in viscosity of the materials. The DuNoUy tensiometer was selected for this purpose. The contention that the bubble pressure method is applicable only in the case of fluids which have a viscosity below a certain maximum value is substantiated by the results of a series of surface tension determinations obtained by the bubble pressure method and a determination on the same tar carried out a t room temperature with the DuNoiiy instrument. The results are shown in Figure 7. Since the same trend was observed in the case of each of the other tars, the bubble pressure method was used between 80" and 90" C., whereas the surface tensions of the materials a t room temperature were determined by the DuNoUy tensiometer. The apparatus used to determine the maximum bubble pressure was built in close analogy to that developed by Fricke and Meyring (4) and is shown diagrammatically in Figure 8:

.

Tlie-tar is brought into sample tube X and immersed in a constant-temperature bath. After the desired tem erature equilibrium has been reached, the glass tube, G, provifed with a fine o ening at the end, is gently lowered until its tip touches the surface of the tar. Contact is readily obtained by use of the screw arrangement, M . Then, the two stopcocks are adjusted so that the mercury in container A flows slowly into container B. Since the air in B is gradually displaced by mercury, pressure is developed in the system. At a definite maximum pressure the air bubble formed on the tip of the glass tube bursts, thereby releasing part of the pressure in the system. This maximum pressure is a direct measure of the surface tension of the liquid under consideration. The theoretically correct formula for the evaluation of surface tension from the maximum bubble pressure was developed by Schrodinger (16) : u = ; x p x

where

5

=

i1 - 5 x - - 2

p X r

1p2Xr2

surface tension

p = maximum ressure = p = T

radius of tRe tip density of the liquid

As the correction introduced by the second and third members, of this equation is of a magn"itude insignificant compared

VOI,. 28, NO. 6

to the accuracy of the present determinations, the formula can be simplified to : r u = - x p 2

Since a n inclined gage was used, the angle of which was not readily ascertained, an over-all constant for the apparatus was determined by measuring the maximum bubble pressure for water, nitrobenzene, and benzene. Each determination consisted of two separate sets of at least five consecutive readings. For each set the capillary was removed from the liquid, cleaned, and brought back into contact with the liquid. The rate of air displacement was regulated to give less than three bubbles a minute. In the case of tars it was found necessary to reduce this rate to less than two bubbles a minute. The following results were obtained in the calibration tests: Max. Bubble Pressure at

25' C. in Cm a8 Obsvd. on''

Substance Water Nitrobenaene Beniene

Inclined Gage 12.71 6.82

3.7s

Surface Tension Taken or Interpolated from Intern. Critical Tables 71.92 43.2 28.22

Using these ~mlues,the following equation for the relation between surface tension and maximum bubble pressure was derived for the apparatus used: u =

where h

=

4.874 X h

+ (3.97

pressure as read from inclined gage, cm.

The accuracy of the method was tested by calculating from the surf ace tension of the reference liquids the corresponding bubble pressures and comparing the computed values with those actually observed. The mean deviation was 0.4 per cent; the maximum deviation, 0.7 per cent. The results obtained from the determinations of the maximum bubble pressures on the various tars are tabulated in conjunction with those from the DuNoiiy determinations. The most reliable values are those obtained a t 90" C. At 80" C . the effect of increased viscosity affected the results, particularly in the case of the unfiltered tars. For this reason the values for the unfiltered tars a t this temperature are omitted. I n the case of tar 9-H which is the highest in suspended matter, R reliable figure could not even be obtained a t 90" C. The determination by the DuNoUy tensiometer method was performed on a standard instrument with B platinum ring of 4-em. periphery. The instrument was calibrated by placing small weights on the ring and determining the load per scale division.

JUNE. 1936

INDUSTRIAL AND ENGINEERING CHEMISTRY

Because of the high viscosity of the materials a t room temperature, the tension on the ring had to be increased very slowly, Usually preliminary tests were performed to find the range of the tension a t which rupture of the film would take place. I n a final test this range would be approached quickly, and the tension a t which deformation to final rupture occurred could be ascertained by successive small increases in the load and careful observations of the deformation of the film adhering to the ring. Each value given in the following table is the average of two determinations. The maximum deviation of any observation from the mean was one per cent. However, the absolute accuracy of the DuNoUy values are probably lower because of instrumental technic. The results of the surface tension determinations are as follows: Sample No.

1-v

Condition of Tar Unfiltered Filtered

Surface Tension, Dynes/Cm. Bubble pressure DuNouy 900 c. 800 c. 32" 31 5 36 9 31 4 32 0 36 3

c.

2-V

Unfiltered Filtered

32.1 32.2

32:9

36.9 36.8

3-c

Unfiltered Filtered

35.7 34.9

36:5

40.9 40.9

4-c

Unfiltered Filtered

37.5 37.3

38:2

46.9 45.3

5-c

Unfiltered Filtered

39.8 39.9

4616

48.0 48.0

6-C

Unfiltered Filtered

40.1 40.2

40: 9

46.3 46.5

7-c

Unfiltered Filtered

39.8 39.9

40:6

46.7 48.6

8-H

Unfiltered Filtered

38.9 38.6

39:4

46.1 45.5

9-H

Unfiltered Filtered

3i:2

38:7

45.9 44.7

The values show that, in general, the difference between the surface tension of a n unfiltered and the corresponding filtered tar is less than one per cent, or smaller than the precision of an individual determination. For the values determined by bubble pressure, this holds true with the exception of tar 3-C. I n the case of the DuNoUy determinations discrepancies up to 4 per cent occur in those tars which have a relatively high concentration of C-I. Since the same differences do not appear in the corresponding results obtained from the bubble pressure test, it is probable that the accuracy of the DuNoUy method decreases with an increased amount of C-I. I n general, therefore, we may say that the suspended phase has no appreciable effect on the surface tension of coal tars. A comparison of the three types of tars shows that the vertical-retort tars have surface tensions lower than those of the two other groups. The values for the coke-oven tars a t 90" C. vary between 35 and 40 dynes per cm. The horizontal-retort tars also fall within this range. The increase of surface tension with decrease in temperature is of interest with respect to the proposed method of Nicholson (26)for the determination of adhesion of tars to aggregate. As mentioned previously, Nicholson assumed from data published by Nellensteijn ( I I ) , that the surface tension of coal tar at, ordinary temperatures is considerably higher than that of water, whereas a t elevated temperatures the opposite is the case. In view of the relatively small increase in surface tension values shown in the previous table between 90" and 32" C., this appears to be unwarranted. It is highly improbable t h a t a sudden increase occurs between 32" C . and the freezing temperature of water which would raise the surface tensions of the tars to values higher than that of water. The break which Nellensteijn observed in the tem-

731

perature-surface tension relation was dne to the interfering effect of high viscosities.

Adhesion to Aggregates In order to study adhesion to aggregate, a series of tests was performed in accordance with the method of Riedel and Weber ( I & ) , previously described. The aggregate used was limestone. Following the suggestions of Riedel and Weber, nine sodium carbonate solutions were prepared. The most concentrated solution had a molarity of 1, the second of the third of 1/4, etc. The last and weakest solution had,a molarity of Preliminary tests showed that, in the case of most tars, complete separation of the binder took place only when the aqueous phase was strongly alkaline. Boiling of these solutions was difficult because of the occurrence of violent bumping. The test could be performed with greater ease if that concentration was taken as the measure of adhesion tension which, during one minute of boiling, brought about sufficient displacenient of the tar to give the aqueous phase a turbid appearance. This end point could easily be ascertained and readily duplicated. It was used in the testing of all tar samples. The results of the tests are given in Figure 9. The abscissa shows the individual samples arranged according to their sample numbers. The lines above each sample number indicate the molarity range of the solutions capable of displacing tar from the limestone. The solid lines represent the unfiltered tars, the dotted lines the filtered samples. I n some cases the critical concentration was between j&o reference solutions. In Figure 9 this was indicated by placing the point representing the critical concentration halfway between that reference solution unable to displace the binder and that capable of removing appreciable quantities of it.

FIGURE9. RESULTS OF WETTINQ TES Solution capable of diaplacjng unfiltered tar f --- Solution capable of displacing filtered tar from q i x

Apparently, the solid phase in tars does not influence the outcome of this test. In six instances the same reference solution was required to displace corresponding unfiltered' and filtered samples. Filtered coke-oven tars 5-C and 7-C adhered somewhat more firmly than the same unfiltered tars. In the case of horizontal-retort tar 9-H there was a noticeable difference in favor of the unfiltered sample.

'

Discussion of Results The results of the experimental work have been presented in tabulated form a t the end of each section. The significance . of each series of investigation has been discussed without reference to the results of other sections. Here a n attempt will be made to correlate the experimental data and to discuss their general meaning. In order to avoid frequent reference to the different parts of this paper the main results have been retabulated. They appear in Table V arranged according to the specific gravities of the unfiltered tars.

INDUST~IAL AND ENGINEEKING CHEMISTRY

732

TABLE V.

Verticalretort

1-V 2-V

Unfiltered Filtered Unfiltered Filtered

By-product 3-C coke-oven 4-C

Unfiltered Filtered Unfiltered Filtered 5-C Unfiltered Filtered 6-C Unfiltered Filtered 7-C Unfiltered Filtered

Hori~ontal- 8-H Unfiltered retort Filtered 9-H TJnfiltered Filtered a

Ring and ball,

%% -?'

Log tz

166.5 160.7 162.4 168.7

1.154 1.145 1.156 1.150

13.7 43.6 39,700,000 13.7 4 0 . 1 14.9 45.7 53,500,000 16.8 44.4

SEMMARY OF RESULTS

Coarse

3.15

2.99

3.6

Coarse

2.55

3.55

3.4

184.6 1.175 16.1 188.0 1.172 1 7 . 1 181.9 1.195 1 5 . 5 187.2 1.193 16.4 182.1 1.207 1 3 . 4 171.2 1.202 1 7 . 0 173.9 1.215 1 0 . 2 167.8 1.209 11.3 183.7 1.218 1 3 . 6 186.2 1.207 1 7 . 3

49.0 49.0 55.2 53.3 52.0 52.2 39.6 40.1 51.4 56.7

184.3 1 . 2 4 1 134.6 1.196 176.3 1.270 110.1 1.202

54.5 107,000,000 Verycoarse 17.60 17.65 23.9 53.7 67.6 281,000,000 Verycoarse 21.67 22.32 28.9 59.4

14.4 18.9 18.9 22.9

19,200,000 Very fine

0.44

1.00

1.0

42,800,000

Very fine

0.65

1.62

1.3

75,200,000

Very fine

114,000,000 Fine 91,200,000

VOL. 28, NO. 6

Fine

2.76

1.91

4.0

2.88

3.37

4.2

5.17

4.71

5.7

5.5

5.44 5.61 3 . 9 3.01 3.12

0.5

3.32 3.36 5.47 5.56 3.8 6 . 1 2 6.24 4.0 4.07 4.21 5.6 6.47 6.79 1.4

24.4 28.2

5.37 6.52 6.63 8.53

7.43 7.66 4.93 5.11

8.97 9.25 5.99 6.21

11.09 11.43 9.04 9.37

5.94

4.88 4.93 9.46 9.62 10.66 10.87 8.99 9.30 11.41 11.97

7.34 7.42 15.66 15.92 21.1 21.5 19.02 19.69 25.2 26.4

12.17 12.30 26.1 26.5 26.5 27.0 26.2 27.1 31.2 32.7

6.53

35.7 34.9 37.6 37.3 6.49 3 9 . 8 39.9 6.50 4 0 . 1 40.2 6.35 3 9 . 8 39.9

7.92 9.62 11.93 15.36

11.62 14.12 20.8 26.8

18.5 22.5 23.1 29.7

6.03

6.05

31.6 31.4 32.1 32.2

6.55

38.9 38.6

5.26 38:2

- log t ,

This type of presentation shows a t once that most of the tabulated properties change uniformly with the rise in specific gravity of the tar samples. This holds true particularly if the three types of tar are considered separately. QUANTITY AND QUALITY OF SUSPENDED MATERIAL.Within the range of the coke-oven and horizontal-retort tars the number of particles and the amount of the suspended solid phase increase with specific gravity. The lightest of the cokeoven tars contains 19,200,000 particles per mg., the heaviest 91,200,000. The amounts of C-I as determined by Hodurek's procedure (using carbon disulfide as solvent) are 1.00 and 4.71 per cent, respectively. The other coke-oven tars follow this trend closely. The only exception is sample 6-C which contains a number of particles per mg. greater than that of the heaviest coke-oven tar. T h a t its character is somewhat different from that of the other coke-oven tars may also be seen from the small auantitv of its constituents boiling below 300' C. The relativelv larie amount of oil which ha> to be removed from its crud; tar-to bring it to a float test of 180 seconds a t 32" C. suggests a similarity between it and the vertical-retort tars. The specific gravities of the two continuous vertical-retort tars are so close together that it is not possible to state that the same trend is apparent in this group of tars. With the exception of the verticalretort tars there appears to be a relationship between specific gravity and fineness of suspended material. The coarser suspended material in the low-gravity vertical-retort tars is probably due to the presence of coal particles. By-product coke ovens and horizontal retorts are operated intermittently, whereas the vertical retorts from which samples V-1 and V-2 were obtained are operated continuously. I n such continuous vertical retorts, tar vapors leave the

carbonization equipment countercurrent to the path of the coal. Fine coal particles may be carried away by the vapor stream to a greater extent than in the case of coke ovens or horizontal retorts. SOLVENT-INSOLUBLE MATERIAL. If the quantity and quality of the dispersed phase is related to specific gravity, there may also be a relationship between the concentration of the substances of high molecular weight which can be precipitated by suitable solvents and specific gravity, That this generally is the case may be seen for each type of carbonization equipment from the increase in the concentration of C-I1 accompanying a rise in the amount of the solid phase. To illustrate this relationship, Figure 10 shows the concentration of C-I, C-11, and the number of particles suspended in a milligram of tar plotted against the specific gravities of the unfiltered tars. To indicate trends, curves were drawn through - the Doints within the limits of each tvve of carbonization equipment. This could not be done ior verticalretort tars since t h e g r a v i t i e s of the two samples are close together. Outside of the gravity ranges of the three types of tars, corresponding curves were connected by dotted lines for convenience. The concentration of C-I1 present in the horizontal-retort tars is somewhat smaller than that in the majority of the coke-oven tars, whereas the amount of C-I is considerably higher. Apparently, the carbonization conditions in horizontal r e t o r t s favor polymerization reactions which result in the formation of those heavy compounds which are present in tars in the form of solids. VISCOSITY INDEX.To indicate B possible relationship between specific gravity and viscosity properties, the temperature coefficient of viscosity is plotted in Figure 11 against t h e specific gravity of the unfiltered tars.

JUNE, 1936

INDUSTRIAL AND ENGINEERING CHEMISTRY

733

Regardless of type FIGURE N the values for all the tars fall on a smooth curve. A maximum appears within the range of the cokeoven tars. On first examination t h i s trend seems peculiar, However, the underlying c a u s e becomes apparent when it is realized that the suspended phase has only a stiffening effect on the liquid phase but does not influence the rate of change of viscosity with temperature. I n other words, the temperature coefficient of viscosity depends solely upon the composition of the liquid phase. I t s lowest values were found in the horizontal-retort tars. The liquid 0 phase of these tars is considerably softer than that of the coke-oven tars, They contain more low-boiling constituents, as may be seen from the distillation analyses of their filtered tars. Possibly these low-boiling constituents act as solvents and prevent the solidification of certain resinous materials over wider temperature ranges. To illustrate these considerations the principal data obtained from the examination of filtered tar 9-H and of the same with specific gravity may be observed within the range of the sample after it had been distilled to the consistency range of the original unfiltered tar are retabulated: vertical-retort and coke-oven tars. However, the trend is reversed in the range of the horizontal-retort tars. T h a t this Distd. and Filtered Filtered change is due solely to the influence of the suspended phase Float test a t 32' C.,8ec. 110.1 215 on the specific gravity may be seen from Figure 13 where Temp. coefficient of viscosity 5.26 6.30 1.202 I.212 Sp. gr. at 26/25' C. the surface tension is plotted against the specific gravities on the filtered tars. The resulting curve is substantially a These results were also incorporated in Figure 11. They show straight line on which the values of all three groups of tars that for a comparison of the inherent viscosity properties fall with a surprising degree of accuracy. of tars their liquid phases must be brought to the same consistADHESIONTO AGGREGATES.To investigate a possible ency. However, with respect to the results of the present relation between adhesion to aggregates and the surface investigation, we may say that the observed differences in tension determinations, Figure 14 shows the molarities of the temperature coefficients of viscosity are due principally the reference solutions capable of displacing the unfiltered to variations in the fluidity of the liquid phases of the tars. samples, plotted against their surface tensions as determined The fluidity, in turn, appears to depend upoii the concentration by bubble pressure a t 90" C. The resulting curve indicates a of low-boiling constituents or, more generally, upon the definite trend. In general, an increase in the surface tension percentage of oils with solvent properties. The differences of a tar is accompanied by an improvement in its adhesional in the temperature coefficient of viscosity are not characteriswetting properties with respect to limestone. tics inherent in the three groups of tars. Since there is a relationship between surface tension and SURFACE TENSION.As indicated before, the surface tension specific gravity of a coal tar, a similar trend may be expected of a tar is independent of the solid phase. It is affected only for the relation between wetting properties and specific by the components of the liquid phase. Therefore, a direct gravity. However, an attempt to plot this relation shows relationship between the specific gravity of a tar and surface that it is not nearly so well defined. tension may be expected only where the specific gravity is not In order to find a possible explanation, it was attempted influenced substantially by the solid phase. In Figure 12 to relate the wetting properties to the amount of GI1 as the surface tensions of the various tars are plotted against determined by various solvents, a relation which has been their specific gravities. A regular increase of surface tension indicated by Demann (1). The most consistent results were found when the molarity of the sodium carbonate solution capable of displacing the tar from the tarlimestone mixture was plotted against the E's D amount of C-I1 insoluble in acetone. This relation is shown in Figure 15. I n order to. 88 eliminate the possible influence of the solid phase in the tars, all values refer to the filtered tar samples. The resulting straight-line relationship indicates that the percentage of binding constituents in coal tars may perhaps be evaluated or indicated by the amount of C-I1 as determined by acetone extraction. However, assurance of the existence of this relation must necessarily depend upon a conMOL4RLV Of NaCO, SOLM/Oh' SUfF/C/€ffT firmation of the assumed significance of the AMOUNT /NJ(1IUU /N ACWVNE m LVSPL~CE u w - f m w nw WPLES M U C€hT BY W N l i n c)c F/LTR€D TM8 Riedel and Weber test procedure. fROH M/X

I

b

$8

t,

INDUSTRIAL AND ENGIKEERING CHEMISTRY

734

Literature Cited (1) Demann, W., Brennst0.f-Chem., 14, 121 (1933). E, V., and Pickard, H., "An Investigation into the Nature and Properties of Coal Tar," South Metropolitan Gas Co., 1931. (3) Eymann, W., Asphalt Teer Strassenbautech., 33,751 (1933). (4) Fricke, R., and Meyring, K., Ibid., 32, 2 6 4 (1932). (5) Green, H., J. Franklin Inst., 192,637 (1921). (6) Herschel, W. H., J. 1x11.ENG.CHEM.,14,715 (1922). (7) Hodurek, R., M i t t . I n s t . Kohlenuergasung, 1, 9 , 19, 2 8 (1919). 181 Klinkmann. G. H.. Doktor-Dissertation. Technische Hochschule, Karlsruhe,' 1931. (9) Nellensteijn, F. J., Teer u. B i t u m e n , 3 1 , 309 (1933). (10) Nellensteijn, F. J., 2. angew. Chem., 43,4 0 2 (1930). (11) Nellensteiin, F. J., and Rodenburg, N. K., Kolloidchem. Bei(2) Evans,

\-,

~~

hefte, 31, 4 3 4 (1930).

VOL. 28, NO. 6

(12) Nicholson, R. N., Roads and Streets, March, April, June, 1932, pp. 136, 165, 237. (13) Oberbach, J., Asphalt Teer Strassenbautech., 33, 8 1 4 (1933). (14) Pochettino, A., N u o v o cimento, [61 8,7 7 (1914).

(15) Riedel, W., and Weber, H., Asphalt Teer Strassenbautech., 33, 677 (1933).

(16) Schrodinger, E., Ann. Physik, [ 4 ] 46,413 (1915). (17) Spilker, A., Asphalt Teer Strassenbautech., 31, 957 (1931). (18) Tietze, W., Ibid., 33, 7 1 8 (1933). (19) Ubbelohde, L., Ullrich Ch., Walter, C., Oel u n d Kohle, 11, 8 6 4 (1935). (20) Westmeyer, R., Asphalt Teer Strassenbautech., 34,2 9 1 (1934). (21) Work, L. T., Proc. Am. SOC.Testing Materials, 28,Pt.I1 (1928). RECEIVBD February 28, 1935. Condensed from a dissertation submitted by Ernest W. Volkmann in partial fulfillment of the requirements for the degree of doctor of philosophy in the Faculty of Pure Science of Columbia University.

SESAME SEED PROTEIN Fat Extractants on Solubility in Salt

and Alkali WILLIAM H. ADOLPH' AND I. LIN Yenching University, Peiping, China

N THE absence of a dairy industry which produces casein, increased attention is being devoted in China to vegetable proteins from the point of view of industrial application as well as of conserving the nation's protein food supply. Considerable attention has already been given to soy-bean protein, the so-called vegetable casein. Sesame seed, however, which contain about 62 per cent oil, also contain 20 to 22 per cent protein. Sesame oil is a n important article of commerce in China, and a t present the protein remaining in the residue after the oil is removed is largely discarded or used as fertilizer. The oil is obtained either by expression with simultaneous heat treatment or by extraction with a fat solvent. Sesame protein, which Jones and Gersdorff show consists essent,ially of globulin, can be extracted from the residual oil-free cake by treatment with sodium chloride solution or other globulin solvents. If the treatment of the sesame seed during the process of removing the oil should result in denaturing the protein, its solubility would be affected accordingly. Sesame seed were submitted to different types of treatment; the results are here reported of the protein solubility in solutions of sodium chloride, hydroxide, and carbonate, and also the yield of protein obtained by different methods of coagulation.

Experimental Procedure SAMPLE. The sesame seed used were of the white variety grown in the Peiping area. The material was washed thoroughly with water and air-dried; after it was ground, the oil 1 Present address, care of School of Medicine, Yale University, New Haven, Conn. 1 J . B i d . Chem., 75, 216 (1927).

Sesame seed contain about 22 per cent vegetable protein, mainly globulin, which has properties similar to "vegetable casein," and may be applied as a plastic and an adhesive. The solubility of the sesame protein in the common protein solvents, sodium chloride, hydroxide, and carbonate, is shown to be not seriously affected by previous treatment with gasoline, or by a temperature of 110" C., while treatment with methanol causes a decided decrease in solubility especially in sodium chloride and sodium carbonate solutions.

was extracted with ether for 24 hours in a Soxhlet apparatus. It was assumed that treatment with ether a t this temperature (not over 35" C . ) does not harm the protein. TREATMENT OF THE OIL-EXTRACTED SEED. The seed from which the oil had been removed was subjected to one of the following methods of treatment: (1) Heat treatment. Forty grams of the sample were placed in an electric oven at 110' C. for 3 hours. ( 2 ) Treatment with gasoline. To 40 grams of the Sam le were added 100 cc. of commercial gasoline (boiling oint, 85" in a stoppered Erlenmeyer flask and kept at 60' for 3 hours, after which the gasoline was removed by evaporation. (3) Treatment with methanol. Same as in (2), substituting methanol for gasoline.

8

8.)

SOLUBILITY.One-gram samples of the material which had been submitted to one of the above treatments were treated with 80 cc. of the given solvent solution and shaken in a thermostat for 3 hours a t 25' C., and nitrogen was determined in 25 cc. of the filtered solution by the Kjeldahl method. Sodium chloride, hydroxide, and carbonate were employed as solvents. The results of these measurements are shown in Figure 1, where the concentration of sodium chloride is expressed in normal units and that of the alkaline solvents in terms of p H value, determined in each case colorimetrically in the resulting protein solution.