V O L U M E 2 2 , NO. 7, J U L Y 1 9 5 0
867 (2) Do=, M. P., “Physical Constants of the Principal Hydrocar-
Table IV.
Range of Physical Properties of Gas Oil Aromatic Typesa
Monocyclic Dicyclic Tricyclic a From data of Doss b Estimated.
n %o 1.48-1.54 1.53-1.60 1.59-166b (2)and Egloff ( 8 ) .
8F.C 125-160 160-260 300-,550
-
3
(n 1.05-1.07 1.09-1.11 >1.1
The monocyclic aromatics of the straight-run oil were divided into two distinct bands. The same was true of the dicyclic aromatics of the catalytically cracked material; this indicated that further type distinctions may derive from the application of this approach. ACKNOWLEDGMENT
The authors wish to acknowledge the advice and cooperation of K. E. Train, R. G. Appleby, hl. 0. Baker, and G. P. Hinds, Jr., in connection with this work.
bons,” 4th ed., New York, Texas Co., 1943. (3) Egloff,G., “Phyeical Properties of Hydrocarbone,” Vols. 3 and 4, New York, Reinhold Publishing Corp., 1946-47. (4) Gooding, R. M., and Hopkins, R. L., paper presented before Division of Petroleum Chemistry, 110th Meeting, AM. CHEM.Soc., Chicago, Ill., 1946. (5) Lipkin, M. R., et al., ANAL.CHEM.,20, 130-4 (1948). (6) Lipkin, M. R., and Martin C. C., Ibid., 19, 183-9 (1947). (7) Lipkin, M. R., Martin, C. C., and Hoffecker, W. A., paper presented before Division of Petroleum Chemistry, 113th Meeting, AM. CHEM.SOC., Chicago, Ill., 1948. (8) Lipkin, M. R., Martin, C. C., and Kurtz, 8. S., IND. ENQ. CHEM.,ANAL.ED.,18,376-80 (1946). (9) Mair, B. J., J . Research Natl. Bur. Standarde., 34, 436-61 (1945). (10) Mair, B.J., and Forriati, A. F., Ibid., 32, 165-83 (1944). (11) Mair, B. J., Gaboriault, A. L., and Rossini, F. D., Znd. Eng. Chem., 39,1672-81 (1947). (12) . , Mair. B. J.. and Rossini. F. D.. A.P.I. Project 6. “ReDort on Fractionation Analysis and ‘Isolation of Hydroca&ns in Petroleum,” Mareh 31, 1947. (13) Mair, B. J., Sweetman, A. J., and Rossini, F. D., Znd. Eng. Chem., 41, 2224-30 (1949). (14) Mills, I. W., Hirschler, A. E., and Kurtz, S. S.. Ibid., 38, 442-50 (1946).
LITERATURE CITED
( I ) Deanesly, R. M., and Carleton, L. T.. IND.ENG.CHEM.,ANAL. En.,14. 220-6 (1942).
RECEIVEDMarch 10, 1950. Presented before the Division of Petroleum Chemistry, Symposium on Adsorption, at the 116th Meeting of the AMmBICAN CHEMICAL SOCIETY, Atlantic City, N. J.
Separation of Nitrogen Compounds by Adsorption from Shale Oil J. R. SMITH, C. R. SMITH, JR., AND G. U. DINNEEN Petroleum and Oil-Shale Experiment Station, Bureau of Mines, Luramie, Wyo. One of the complicating factors in elucidating the composition of distillates from Colorado shale oil is the presence of large quantities of nitrogen compounds. The content of these compounds may be as much as 50 weight % on some fractions. A procedure is described for separating shale-oil distillates into two fractions-one containing primarily hy-
M
OSTofthedistillateobtainedfromColoradoshatleoilisin the boiling range above 400’ F. Consequently, knowledge of the composition of this higher boiling fraction is of paramount importance in developing processes for producing engine and other distillate fuels from shale oil. The majority of the methods ( 2 , 6 , 1 1 ) for the analysis of petroleum distillates in the higher boiling ranges estimate the carbon atoms as paraffinic, naphthenic, or aromatic. These methods are based on correlations that usually are not valid for samples containing olefinic unsaturation and nonhydrocarbon material. If the method utilizes optical properties, it is difficult, if not actually impossible to make the required determinations on dark samples. Consequeiitly, higher boiling shale-oil distillates, which always contain large quantities of olefins and of nitrogen compounds and are dark, cannot be analyzed satisfactorily by direct application of available methods. Separation and recovery of the nitrogen compounds, therefore, would aid materially the analysis of the residual hydrocarbons as well as analysis of the nitrogen compounds themselves. A large part of the nitrogen compounds in shale-oil naphthas and kerosene8 can be removed by treatment with dilute aqueous acid (I ), However, this procedure is ineffective for distillates having a 50% boiling point above approximately 525” F., al-
drocarbons and the other primarily nitrogen and other heterocyclic compounds. This separation is based on adsorption and employs Florisil, a synthetic magnesium silicate, as the adsorbent. Factors investigated in selecting optimum conditions for the process are discussed. Reaults obtained on several shale-oil distillates are presented.
though such distillates may contain 30 to 50 weight 70 nitrogen compounds. Adsorption had proved effective in analyzing shaleoil naphtha (8), so an attempt was made to apply the technique to the separation of nitrogen compounds from higher boiling fractions. After preliminary tests on a number of adsorbents, Florisil was selected the best available for the desired application. Conditions were established for a procedure that would separate a shale-oil distillate into two fractions-a lightrcolored one containing essentially hydrocarbons and a dark one containing principally nitrogen compounds. Results obtained by application of the procedure to several samples are presented. APPARATUS
Adsorbent. Florisil, a synthetic magnesium silicate manufactured by the Floridin Company. Air-dried material, 30- to 60mesh,wiw used. Column. A glass tube, similar to that described by Mair and White (8), having a fritted disk or other device near the lower end for supporting the adsorbent end having a reservoir of approximately 200-ml. ca acity on the top. If necessary for some applications, the corumn may be equipped with a heating or cooling jacket and a connection for a plying pressure by means of an inert gas. In the work reporteam this paper Florisil was em-
ANALYTICAL CHEMISTRY
868 ployed only to adsorb nitrogen compounds of unknown structure from distillates obtained from crude shale oil produced in an N-T-Uretort. Consequently, its activity for removing nitrogen compounds from other materials cannot be accurately estimated. Therefore, the quantity of Florisil and size of column necessary for a particular a plication will have to be determined empirically. For N-ZU distillates, Florisil equal to 300 times the weight of nitrogen in the sample and columns having diameters between 17 and 40 mm. and having height to diameter ratios between 12 and 150 have given comparable and satisfactory results. For smaller scale work a column having a diameter of less than 17 mm. may give more efficient separations. Refractometer. An instrument capable of measuring the refractive index to *0.0001. Stripping Unit. E uipment capable of quantitatively removing pentane and methan3 from the fractions. The design and size of this e uipment will depend on the quantity of material being treatel. In this work a batch-type stripping unit consisting of a flask and inverted Friedrichs condenser was used. Pentane. Any mixture of saturated hydrocarbons boiling in the vicinity of 100' F. and having no high-boiling residue. Methanol. Absolute. PROCEDURE
A column of appropriate size is filled with Florisil and packed by tapping. An accurately weighed sample is introduced into the top of the column. It is often convenient to dilute samples that are solid a t room temperature with pentane to prevent solidification in the adsorbent. When the sample has entered the Florisil. 100 ml. of pentane are added to the reservoir a t the top of the column. The pentane and the portion of the sample eluted from the adsorbent are collected in a suitable receiver as they emerge from the column. Successive 100-ml. quantities of pentane are added to the reservoir and allowed to percolate through the column until the pentane issuing from the bottom of the column has a refractive index within 0.0002 of that of the pentane being introduced into the reservoir. Approximately 2000 ml. of pentane are required per 100 grams of sample for the type of material usually se arated in this laboratory. When t i e refractive index of the pentane-eluted fraction has reached the desired value, the receiver under the column is exchanged for a clean one, and methanol is introduced into the reservoir. Successive 100-ml. ortions of methanol are added and allowed to percolate througE the column until the material issuing from the column has a refractive index within 0.0002 of the value for the methanol being used. Approximately 1200 ml. of methanol per 100 grams of sample are usually required. The pentane and methanol are stripped from their respective fractions in tared flasks. When the tared flasks have reached essentially constant wei ht, the quantity of material eluted by pentane and that eluted%ymethanol are calculated. Nitrogen is determined in each of these fractions by a modification (un ublished) of the Kjeldahl macroprocedure to check the extent ofthe separation of nitrogen compounds that has been achieved. The pentane-eluted material usually will be light-colored, color-stable, and ractically free of nitrogen, while the methanol-eluted materiafwill be dark and high in nitrogen. DISCUSS O N
It is theoretically possible to use as an adsorbent any solid material that will differentially adsorb the desired components from the sample and will not decompose the constituents of the sample. However, efficient application of the adsorption technique requires in addition to selection of an active adsorbent consideration of a number of additional factors, such as activation and particle size of the adsorbent. An adsorption procedure involving elution with solvents of increasing polarity and similar to one used on petroleum distillates ( 6 , 7 ) was employed for the work reported in this paper. This technique is advantageous for use on high-boiling materials, particularly when it is desired to recover the separated fractions for further investigation. Selection of an Adsorbent. The characteristics desirable in an adsorbent for separating the nitrogen compounds present in shale-oil distillates are: selective adsorption toward the nitrogen compounds; adsorption reversible enough so that desorption of the nitrogen compounds will be a simple operation; sufficient capacity so that a relatively small adsorbent-sample ratio may be used; low selective adsorption toward hydrocarbons so these may be separated from the nitrogen compounds as one fraction; and a
particle size that will give a rapid filtration rate. In practice, of course, the adsorbent selected will represent a compromise of thcse characteristics. Several available adsorbents were compared &s to their adsorption characteristics for the nitrogen compounds in shaleoil distillates For a complete comparison, it would be necessary to obtain the isotherms for these compounds on the adsorbents. However, it was felt that for the work reported in this paper a comparison at a nitrogen concentration representative of that usually obtained on high-boiling shale-oil distillates would give satisfactory results. Therefore, the adsorbents were tested using a distillate having a 50$& boiling point of about 725' F. and a nitrogen content of 1.78 weight yo. The test method was similar to the recommended procedure, except that the adsorbent-sample ratio was reduced so that only about half the nitrogen charged to the column was adsorbed and some of the effluent contained about as high a concentration of nitrogen as the charge. The quantity of nitrogen adsorbed during the pentane elution serves as a qualitative measure of the efficiency of the adsorbent in the desired process. Results for quantity of nitrogen adsorbed and total recovery of nitrogen and samp,e as determined by this overloading test method are given in Table I.
n
0.800
!I-
2 0.mo E
e
0
2 0.400
t
1
0.100
Qooo 0
$00
800 1000 la00 ACTIVATION TEYPERATURL,@f,
moo
Figure 1. Effect of Activation Temperature on Adsorption of Nitrogen Compounds by Florisil
From the standpoint of the quantity of nitrogen adsorbed, color of pentane-eluted fraction, total recovery of sample and nitrogen, and rate of percolation, the Florisil was the best of the adsorbents tested. However, if these desirable characteristics are accompanied by a high selective adsorption for one or more groups of hydrocarbons, the Florisil cannot be used conveniently for separating nitrogen compounds from hydrocarbons. Investigation of this phase of the problem by a series of experiments indicated that, using Florisil, it should be possible to remove most hydrocarbons from nitrogen compounds by an elution procedure. Preliminary work on the preceding problem showed that aromatic hydrocarbons are somewhat more strongly adsorbed, by Florisil than saturates or olefins. Consequently, the separation of
869
V O L U M E 2 2 , NO. 7, J U L Y 1 9 5 0 Table I.
Adsorbent Silica gel Alumina Florex Florite Florisil Carbon
Adsorption of Nitrogen Compounds by Various Adsorbents Total Matenal Recovered
Nitrogen Recovered
Nitrogen Retained during Pentane Elution
%
%
G./Z00 0.
99.6 98.8 98; 9
94 93 88
99.6
95
0.39 0.41 0.40 0.37 0.65
6
No percolation of methanol could be obtained. b Carbon available had such a slow percolation rate that no results were obtained.
aromatics from nitrogen compounds was studied. Because of ready availability and ease of running, some work was done using pyridine and hydrocarbons in the naphtha-boiling range. However, operating variables, particularly activation of the adsorbent, had much greater effects on the adsorption of pyridine than on that of the nitrogen compounds in higher boiling shale-oil fractions. It appeared, therefore, that results obtained using pyridine could not be considered typical. It was found that under adsorption conditions similar to those used on shale-oil distillates nearly quantitative elution of 1methylnaphthalene, 1, 2, 3, 4tetrahydronaphthalene, and 1, 3, 5triethylbenzene was obtained. A synthetic mixture was prepared which contained about 60% of a nitrogen-compound concentrate prepared from a shale-oil fraction by adsorption on Florisil and about 13% of each of the hydrocarbons mentioned in the previous sentence. Placing this mixture in a column containing Florid and eluting with pentane permitted separation and recovery of the hydrocarbons. It was also found that anthracene and phenanthrene could be eluted from Florisil by pentane if they were dissolved in benzene before being placed on the adsorbent. Effect of Activation Temperature. Florisil used in some of the preliminary work had been activated at 1200" F. by the manufacturer and had not received any further treatment in this laboratory. To determine the effect of activation temperature, a quantity of Florisil was obtained that had been air-dried only by the manufacturer. Samples of this material were then activated in this laboratory by heating for 4 hours a t one of several temperatures between 300" and 1600" F. The activity of these heated samples waa determined by the overloading test method described previously. The results obtained are shown in Figure 1. It is evident that over a considerable range the activation temperature for the Florisil has relatively little effect on the adsorp tion of nitrogen compounds. However, the adsorption capacity of the material decreases very rapidly if a temperature above 1200"F. is used. Although activation of Florisil by heat increases ita adsorption rapacity for nitrogen compounds, the activation causes an even greater relative increase in its adsorption capacity for some nonnitrogen containing compounds in certain shale-oil fractions. Consequently, unactivated, air-dried Florisil is usually used for the separation of nitrogen compounds from shale-oil distillates by the procedure described in this paper. However, as activation apparently alters the adsorption characteristics of the Florisil, proper choice of activation conditions and eluents should make Florisil useful for a variety of separations. Particle Size of Adsorbent. Decreasing the particle size of an adsorbent usually will increase its adsorption capacity (f 0), but will decrease the percolation rate. Therefore, it is advantageous to select the largest particle size that will perform the desired separation. Florisil is furnished by the manufacturer in three particle-size grades, designated by mesh as 30 to 60, 60 to 100, and 100 up. An air-dried sample of each of these was activated for 4 hours a t 500' F. and then evaluated by the overloading test method. The
results in Table I1 indicate that, when all conditions except particle size are constant, the capacities of the three grades of adsorbent are practically the same, but the percolation rate through the 30- to 60-mesh material is faster. Consequently, use of this material is recommended. APPLICATION TO SHALE-OIL DISTILLATES
Four shale-oil distillates obtained from oil produced in an N-T-U retort were treated by the recommended procedure. The results are given in Table 111. The average molecular weights were calculated by the method of Mills et al. (9). The content of nitrogen and sulfur compounds in each fraction was calculated assuming only one atom other than carbon and hydrogen to the molecule. It is apparent from the results in Table I11 that the ot@! recovery of sample is good; that well over 90% of the nitrogen recovered in the process is in the methanol-eluted material; and that separation of the sulfur compounds into either of the fractions is never attained. As the solvent stripping procedure employed was designed primarily for use on higher boiling fractions, the recoveries on the naphtha are poor and modifications would have to be made in the technique to adapt it for convenient use on this material.
Table 11.
Effect of Particle Size of Florisil on Adsorption of Nitrogen Compounds Nitrogen Retained durin Pentane dilution loo 0. 0.66 0.65 0.66
Particle Size, Mesh 30-60 60-100 100 UP
Approximate Time Required for Pentane Elution Houra 2.5 3.25 4.0
Considering the large quantity of nitrogen compounds in the methanol-eluted fraction on the higher boiling materials and that this fraction contains substantial quantities of sulfur compounds and probably most of the oxygen compounds, it would seem that there are few if any hydrocarbons in the fraction.
1.500
1.460
i
I
I
I 1.4
00
I
I
I
I
I
I
ID
20
10
4.0
51)
6D
I 70
8,O
FILTRATE, gram8 Adsorption Analysis of Pentane-Eluted Figure 2. Fraction from N-T-U Heavy Gas Oil
BD
WlOHT OF
Although the complexity of the original sample makes a direct analysis extremely difficult, the pentane-eluted fraction contains mostly hydrocarbons, so an estimate of its composition may be obtained by available methods. Results of analysis of the pentaneeluted fraction from the heavy gas-oil sample are shown in Figure 2. The aromatics were obtained by the elution technique of Lipkin et al. ( 5 ) . By this method any molecule containing one or
ANALYTICAL CHEMISTRY
870
Table 111. Separation of Nitrogen Compounds from Four Shale-Oil Distillates by Use of Florisil Av. Mol. Wt.
Material Recovered, in Fraction
wt.%
Nitrogen
wt.%
Nitrogen Compounds
Relative Quantity of Recovered Nitrogen in. Fraction
wt.%
%
Sulfur Compounds
Relative Quantity of Recovered Sulfur in Fraction
Wt.%
wt.%
%
0.65
6.5
..
0.63 0.67
6.3 6.7
55.9 44.1
Sulfur
N-T-U" Heavy Gas Oil, Boiling above 625' F. Original sample Fractions Pentane eluted Methanol eluted
320
Ori ins1 sample &actions Pentane eluted Methanol eluted
280
Original sample Fractions Pentane eluted Methanol eluted
200
Original sample Fractions Pentane eluted Methanol eluted
140
55.5 41.3
1.78
40.8
0.11 3.73
2.5 85.4
3.7 96.3
N-T-U Light Gas Oil, Boiling Range 570-750° F 62.6 37.0
1.82
36.4
0.13 4.34
2.6 86.8
4.7 95.3
0.71
6.2
..
0.81 0.56
7.1 4.9
71.0 29.0
N-T-U Kerosene. Boiling Range 400-585' F. 76.1 23.5
1.41
20.2
0.11 5.08
1.6 72.6
6.4 93.6
0.74
4.7
,.
0.76 0.38
4.8 2.4
86.5 13.5
N-T-U Naphtha, Boiling Range 155-415" E". 1.21 75.0 18.0
" Aninternally fired retort employed
0.0
4 04
12.1
0.0
40.4
0.0 100
1.15
5.0
..
1.08 1.39
4.7 6.0
76.4 23.9
by the Bureau of Mine8 in experimental operations a t the Oil-Shale Demonstration Plant, Rifle, Colo.
more benzene rings is classified as an aromatic. The paraffinnaphthene-olefin fraction obtained by Lipkin's method was rerun by a displacement technique developed in this laboratory ( 4 ) . The displacement technique was not applied directly to the fraction because a suitable desorbent for the aromatics in this boiling range was not available. The results indicate that the composition of the hydrocarbon fraction is approximately 2670 paraffins and naphthenes, 30% olefins, and 44% aromatics and sulfur compounds. SUMMARY
Shale-oil distillates in the boiling range above 400' F. contain large amounts of nttrogen compounds, which complicate efforts to determine the composition of these distillates. Adsorption was selected as a technique which might permit the separation of these distillates into less complex fractions more suitable for subsequent analysis. A procedure is described for separating shale-oil distillates into two fractions. One of these fractions consists predominantly of hydrocarbons, whereas the other contains predominantly nitrogen compounds, The separation is based on selective adsorption of the nitrogen compounds by Florisil, a synthetic magnesium silicate. Results obtained on four shale-oil distillates by the recommended procedure indicate that approximately 95% of the nitrogen will be concentrated in one fraction and that this fraction will consist predominantly of nonhydrocarbons. The other fraction obtained will be a light-colored, color-stable material consisting predominantly of a mixture of saturated, olefinic, and aromatic h:rdrocarbons. ACKNOWLEDGMENT
This work was done under the Synthetic Liquid Fuels program authorized by Public Law 290, iSth Congress, and under the
general supervision of W. C. Schroeder, chief of the office of Synthetic Liquid Fuels. The authors wish to acknowledge the general guidance of R. A. Cattell, chief of the Oil-Shale Research and Demonstration Plant Branch, H. P. Rue, supervising engineer of the Petroleum and Oil-Shale Experiment Station, and H. M. Thorne, engineer-in-charge of oil-shale research and development. J. S. Ball, refinery engineer, was in direct charge of the investige tion. The authors wish to thank Shirley Nix and P. P. Venesiano, who performed part of the laboratory work. This work was done under a cooperative agreement between the Bureau of Mines, United S t a t w Lh7nartment of the Interior, and the University of Wyoming. LITERATURE CITED
(1) Ball, J. S., Dinneen, G. U., Smith, J. R., Bailey, C. W., and Van Meter, R., Ind. Eng. Chem., 41,581 (1949). (2) Deanesly, R. M., and Carleton, L. T., IND.ENQ.CHEM.,*%NAL. ED., 14,220(1942). (3) Dinneen, G. U.,Bailey, C. W., Smith, J. R., and Ball, .J. S., ANAL.CHEM.,19,992 (1947). (4) Dinneen, G.U., Thompson, C. J., Smith, J. R., and Ball, J. 5., Ibid.. 22. 871 (1950). (5) Lipkin, 'M.'R., Hoffecker, W. A , , Martin, C. C., and Ledley, R. E.,Ibid., 20, 130 (1948). (6) Lipkin, M. R.,and Martin, C. C., Ibid., 19,183 (1947). (7) Mair, B. J., and Forziati, A. F., J . Research Nutl. Bur. Standards, 32,165 (1944). (8) Mair, B. J., and White, J. D., Ibid., 15,51 (1935). (9) . . Mills. I. W.. Hirschler. A. E., and Kurtz, S.S.,Jr., Ind. Bng. Chem., 38, 442 (1946). (10) Strain, H.H., "Chromatographic Adsorption Analysis," p. 51, New York, Interscience Publishers, 1942. (11) Vlugtor, J. C., Waterman, H. I., and Van Westen, H. A., J . Inst. Petroleum Teehnol., 18,735 (1932);21,661 (1935). RECEIVED December 27, 1949. Presented before the Division of Petroleum Chemistry, Sympoeium on Adsorption, a t the 116th Meeting of the AMERIC A N CAEMICAL SOCIETY,Atlantic City, N. J.
