ing CARCINOGENICITY STUDIES W. A. DIETZ, W. €I. KING, JR., W. PXIESTLEY, J H . , AND JOHN RENNER, JR. Esso Laboratories, Standard O i l Development Co., Linden, . C . J .
HERE is considerable eridciic,c t , h t certain petroleum fractions can cause skin cancer in man as well as in lower animals conditions of frequent antl prolonged exposure or application. 9 comprehensive survey by Hueper (8) discloses that virtually all of this evidence applies t o products t h a t were produced before the advent of modern catalytic refining processes. Also, much of the older information suffers from inadequat(2 technological description of the materials investigated. For example, the literature is replctcx with experimental and occupational t,unior frequency data for so-called mineral oil, petroleurn oil, naphtha oil, Yhale oil, and 1va.y. Terms like these are quite indefinite with respect to modern refining technology and practice, and pertain in some instances to products that scarcely appear on the modern scene. Although there is no direct evidence a t present that modern high boiling catalytically cracked petroleum fractions have caused cancer in man, experiments have shown ( I 1 ) that some of t.httse fractions can produce a high incidence of tumors or cancers in certain lower animals, especially experimental mice. The tfcrm “carcinogenic” as used in this paper applies t o those materials which produced a significant tumor and/or cancer response \Then repeatedly applied t o thc skin of cei.tain experimental mivc under specified conditiolis ( 1 1j. Early recognition of the potential hazard in some a i thew modern producb led t o an ester erimental program a-hich has been carried out during the p (le in these and associated laboratories. Many virgin stoc been examined, but, particular emphasis was directed toward fractions and blends containing high boiling catalyt,irally cracked materials. Background information comprising parts of this program has rr:cently been published on the distillation properties of polyiiuclear aromatic hydrocarbons ( I ) , on various analytical methods for characterizing such hydrocarbon types in complex petrolcum mixtures (IS), and on some pl,elirninary correlations of thew characterization values with the hiological response data ( 6 ) . The method devised for the qualltitative representation of the biological data h a also been reported (a), as well as the cxperimental and pathological details of the animal tests ( 1 1 j. .I hygiene cont,rol program which \viis 1)iiscd on t,hese findings :tnd is now employed Rnlong refinery workers exposed to potentially hazardous materials has been outlined ( 7 ) . The present’ paper reports further quantitative aspects of thc carcinogenicity of certain refinery streanis, especially a? related to processing antl heavy fuel blending. QUANTITATIVE REPH
KTATJOS
OF THE BIOLOGICAL
Frequent refcrence will be niatlo to the t,unior potency (T.P.) values of various samp1e.l. The method of computing t,hese values a,nd their probable error has been described in detail (a). Briefly, tumor response data are obtained on repeated skin painting of a colony of 30 experimcrital mice with a sample, under specified conditions (11 ). Per cent tumor response values, corrected for nontumor mortdity, are plot,ted against log t,inie on
logarithmic probability paper, and the time in days required for SO% tumor response, t 5 0 , is read from the plot. The tumor potency value is defined as lO,OOO/tj,, and represents a rate of tumor lormation. I n computing the data, no differentiation is made between benign and malignant tumors. The reproducibility of tumor potency values is roughly 110 t o 15 units, depending mainly on the degree of extrapolation used in the determination of tso. Tumor potency values of about 20 or less have doubtful quantitative significance and a relatively large probable error, since they are based on extrapolations from exceedingly low tumor response levels and are comparable t o values sometimes obtained with cont’rol colonies. For example, a tumor pot,ericy value of 20 corresponds t o a t 5 0 value of 500 days, at irhich time t’he number of surviving animals may not exceed 10% of the original colony. It is occasionally convenient, t o describe potencies as low, moderate, or high, even though such cilassifications are quite arbitrary. I n this work, tumor potency values of 50 or 60 or more are regarded as high, 30 or 40 are moderate, and values of about’ 20 are low, or marginal. iilso, while the tumor potency values include both benign and inaligriant t,umors, it has been ohserved t h a t low potency values nearly ;ilways result, from benign tumors only. Wit,h further regard to the significance of a low tumor potency value, this quantity call kc considered as an intrinsic tuniorproducing capacity of a material under the experimental conditions used in the animal t e s t P . However, a variety of evidence in the literature indicates that the ultimate production of tumors also depends on a time factor. T h a t is, under otherwise equal ronditions, a material having a low tumor potency value may produce a higher final tumor response on frequent and prolonged contact or application thaii would a, material of high tumor potency, but applied infrequently or for a short.er period. It does not appear possible a t piescnt t o assign relative weights of importance t o these two factors or t,o determine the manner in which they should be combined into an over-all index of potential hazard. A41so,it is not clear how mouse test results can be translated into the human equivalent, or if indeed a close parallel cven exists. It muet be assumed a t this time t h a t with other things being equal, t.he lowcr the tumor potency value as derived from mouse tests, the l o m n the potential hazard in human esposurP. SLLCK W4XES
Hack waxes are m-ax-oil mixtures corrimonly containing from 15 t o 25% of oil and having vonipositions between t h a t of the crude oils from which they are derived and that of the refined solid paraffin waxes. Slack ivaxes were investigated because it has long been known ( I d ) that sonir virgin crudes have been carcinogenic t o exprriniental mice. Also, it is well known (9) that wax workers with a history of frequent and prolonged exposure qometimes developed occupational skin tumors and cancers (“paraffin cancer”). Table I contains test results for a series 1818
August 1952
I N D U S T R I A L A N D E N G I N E E R I N G CHEMISTRY TARLE I
*ample NO.
142-27 141-27 143-32 144-27 145-27 146-27 147-27 148-32 160-27 149-27 131-32
TUMOR POTENCIES
L)e,ciiption Panhandle slack wax froin first pressing a t 70' F. Panhandle slack wax from second presang a t 22' F. Production pioportion hlend of 142 and 141 San Joaquin slack wax from first pressing at 85" r San Joaquin slack wax from second pressing a t 62' F. San Joaquin slack wax fiom third pressing itt 62O F San Joaquin slack wax fioin fourth pressing a t 41' F. Production propoi tion blend of 144, 145, 146, and 147 Slack wax from hrst pressing of East Texas paraffinic distillate at 64" r Slack wax f r o m second pressing of Ela-t Texas paraffinic distillate a t 36' E' Production proportion blend of 149 and 150
Tumor Potency 14
16
AND INSPECTION D A T A FOR
Gravity OAPI 39 8
io
8
Viscosity, 6SU a t 100' F 76 0
SL4CK Jv4XES
Average Vol. Mol. % ' at Wt 700' F.+ 386 91
48 0
362
1819
Vol
% Aromatics 7
Vacuum Distillation, F., a t Vol. % 5% 50% 95% 641 813 878
82
6
051
760
868
370
0
40 0
19
39 8
73 0
364
87
9
634
810
882
31
38 6
66 0
343
81
10
623
755
868
15
41 6
64 0
343
90
G
680
748
860
20
41 1
58 0
345
65
7
638
717
8.54
17
40 6
347
..
..
..
13
38 4
72 0
359
88
12
668
701
15
38 9
64 0
323
75
9
BR7
736
0
38 7
46 0
333
of refinery slack waxes pressed from several sources of crudes, together with pertinent inspection data for the samples. Although t h e crude source is named under the description of each sample, it is not believed t o have any particular Significance; even crude oils from neighboring locations of a given field can differ widely in their properties. The third column of the table shows t h a t the slack waxes have tumor potency values of less than about 30 and most values are less than 20. It has been mentioned that tumor potency values in the neighborhood of 20 or less have doubtful quantitative significance and a relatively large probable error. This is also evident from the fact that Ihe tumor potency values shown in Table I for production proportion blends (samples 143,148,and 151)do not always agree closely with the corresponding values for the component waxes. It is also apparent that the trend of tumor potency values in a given sequence of wax pressings is occasionally erratic-for example, sample 145, The apparent discrepancies are therefore attributable t o the inherent lack of precision in the biological testing of low potency materials rather than t o some effect arising from the chilling, pressing, or blending operations. It can be concluded from Table I thak there are no significant differences between the tumor potency values of a series of waxes pressed from a given crude, nor are these values appreciably different for slack waxes from the different crudes examined. Also, the inspection data for the waxes, although descriptive, appear t o be of little value for correlation with tumor potency values. Finally, all of the slack waxes examined (with one possible exception) showed very low or marginal tumor potency values. Table I1 shows t h a t the tumor-producing components of the slack waxes are associated with the aromatics fractions. The latter were isolated by silica-gel percolation of the samples. They are seen t o have considerably higher tumor potency values than do the corresponding whole waxes from which they were derived. Other evidence t o be presented later in this paper shows quite clearly t h a t in no case has i t been possible t o find carcinogenicity in any completely nonaromatic fraction of a potent oil. Thus, the nonaromatic components of slack waxes appear t o act as diluents. This is supported by the fact t h a t nonaromatic products, such as U.S.P. petrolatum or U.S.P. white oil, which are included in Table I1 for purposes of comparison, were entirely without effect in the animal tests. [Although Table I1 shows t h a t the nonaromatic compounds
843
"dilute" the carcinogenicity of t h e aromatics fraction of the slack wax, the magnitude of the effect cannot yet be interpreted with respect t o the potential hazard toward exposed workers. The reason is t h a t t h e tumor potency values of the slack waxes are based on tests in which t h e cr7hole wax was applied t o the skin of the mice. Wax workers have often experienced lesions on parts of t h e body (especially the scrotal region) that are protected by clothing. I n such instances, the exposure may have been primarily t o the oil from a crude wax, which could readily penetrate clothing, rather than t o the waxy components proper, which would tend t o be filtered out by the clothing. The testing of slack waxcs and the interpretation of the results in terms of potential hazard are therefore not yet clearly resolved and may constitute a special case.]
TABLETI. TuMon POTEXCIE:~ OF SLACK WAXESAXD THEIR AROMATICS Fn 4CTroNS Slack Wax Oil i n wax, Tumor No.& % Potency 142-27 21 14 141-27 16 19 143-32 20 0 144-27 21 19 145-27 31 25 146-27 12 15 147-27 17 20 148-32 17 20 150-27 15 29 149-27 15 29 151-32 29 0 277-396 .. 0 108-33 e 0 See Table I for descriptions of slack waxes. Low melting U.S.P. petrolatum .41ba U.S.P. white oil. Sample
..
'
-4roinatics Fraction of Slack Wax _ _ Per cent on Tumor slack wax potency 22 30
;
..
12 13 8 9
46 50 31 47
12 11
50 39
.. ..
.. ..
The extensive refining t o which these products, and the closely related highly refined paraffin waxes, are subjected renders them innocuous by removing t,he aromatic components. LUBRICATING OIL FRACTIONS
Various lubricating oil stocks and fractions have been examined, and some of the results are given in Table 111. Most of these stocks are seen to be rompletely noncarcinogenic, although low tumor potency values were found in three cases. Conventional
1820
INDUSTRIAL AND ENGINEERING CHEMISTRY
Vol. 44, No. 8
TABLE 111. TUMOR POTENCIES AND INSPECTION DATAFOR LUBRICATING OIL FRACTIONS
b
Sample NO. Description 107-33 106-32 347-46 354-46 353-46 Phenol extract from Coastal distillate 356-46 Phenol extract from Tia Juana distillate 246-41 Acid-treated paraffinic distillate 348-46 Acid-treated paraffinic distillate 345-46 Phenol extract from Panhandle bright stock 350-46 Determined by separation on a siliaa-gel column. At 210° F.
Tumor Potency 28 31 0 0 0 24 0 0 0 0
acid treating appears t o have no decisive effect on carcinogenicity; two of the acid-treated stocks in Table 111 have fairly low but nonvanishing tumor potency values, whereas three other acid-treated stocks have zero tumor potency values. (Similar acid treatments of a highly carcinogenic clarified oil from catalytic cracking likewise had no appreciable effect on the tumor potency value.) Two of the three phenol extracts were found t o have zero tumor potency values. This may simply mean t h a t the stocks from which these aromatic concentrates were prepared, although they were not tested, were probably completely innocuous. Unfortunately, the stock from which the third phenol extract (sample 356) was prepared was also not tested. However, the low tumor potency value of this extract indicates that its corresponding parent stock very likely would have shown no response in the mouse tests. Since specific polynuclear aromatic hydrocarbons appear to be mainly if not entirelv responsible for the carcinogenicity of high boiling catalytically cracked stocks, as will be pointed out later, the analytical data in Table I11 show, in apparent contrast, t h a t the amount and refractive index of the aromatics present in a lubricating oil stock are without value for predicting its tumor potency value. A plausible explanation is that virgin stocks do not contain appreciable amounts of the highly condensed ring species t h a t are produced in catalytic cracking. Also, the polynuclear aromatic species present in virgin stocks possess rather long alkyl side chaina, the size of which are greatly reduced under cracking conditions. Long side chains are known t o be capable of deactivating carcinogenic aromatic hydrocarbonP ( 1).
Gravity, 0 API 24.0 22.0 17.5 22.0 25.0 I
.
.
...
27.5 26.6 *..
Viscosity S S U a t iooo'F. 109 515 6600 500 515
Bromatics
wt.%a
n%O -
37.2 40.2
1.5627 1 6645
33.2 20.0
1.'i632 1.5274
71.5 29.0 32.8 67.4
1 .'5951 1.5526 1 5496 1.5623
...
...
137b 526 b 110 204 9500
zero or marginally low tumor potency values, but special tests would have to hecarried out on particular samples t o show whethei products of this kind owe their activity (if any) t o the petroleum base or t o the additives. VIRGIN GAS OILS ABD CRUDE RESIDUA
Tests on virgin gas oils are of interest in determining whether the carcinogenicity of certain high boiling catalytically cracked fractions is due primarily t o the cracking process or t o unchanged components from the feed stream. Table I V gives limited data for virgin feeds t o catalytic cracking and also includes data for crude residua, which are used in heavy fuel blends. Thc g s oils were found t o be moderatelv active, the magnitude of theii tumor potency values being about what would be expected if thc crudes from which they were distilled possessed a low ordcr of carcinogenic potency associated with the higher boiling frart ions. The crude residua of Table IT- were found t o have zero tumor potency values. An explanation for the diffeience in the tumor potencj wlue. of the two types of virgin streams shown in Table TV is not ohviouf from the inspection data. The aromatics contents of thc tour samples do not differ much. The refractive indices of thc aromatics from the residua are in fact higher than those from tlie moderately potent gap oils The average molecular weights of thc residua and their content of high boiling species are also higher than the corresponding values for the gas oils. These differrnces could be taken (probably incorrectly) t o mean that the residua have higher contents of polynuclear aromatics than do thc gaq oils and that the zero tumor potency values of the residua are
T.\BI,E1s'. TUMOR POTENCIES AND IKSPECTION DATAFOR VIRGINGAS OILSAND CRUDERESIDUA Sample
Description 361-12 Virgin feed to catalytic cracking 366-42 Virgin feed to catalytic cracking 114-41 Residuum from West Texas crude 175-26 Residuum from Gulf Coast crude a Determined by separation on,a gel column. SO.
Tumor Gravity, Potency OAPI 33 28.4 32 29.0 0 10 8 0 21.4
Visoosity, SSU at
looo F. 62.9 64.0 140.000 3,000
The limited viscosity data of Table 111, and especially some more Pvstematic data to be presented later in connection with heavy fuel blends, have suggested the possibility that viscosity may be an important factor in determining the cai cinogeriic potency of an oil. The idea qeems reasonable because of some evidence ( I O ) that penetration through the pores of the skin or into the hair follicles may be an important condition for the development of the biological response. Work now in progwsq is aimed a t testing this possibility. Finished lubricating oil pioducta, such as coninieicial motor oils, gear oils, and cutting oils, are not included in this discussion because such products are commonly blends of various base stocks and often contain several different and undisclosed chemical additives. A few tests of such products have given cither
Average vel. Mol. % at Wt. 7 0 0 ° F . + 274 45 281 48 793 100 400 82
Vacuum Distillation, F., at Vol. % 5% 50% 95% 496 676 923 520 692 920 879 574 880
. ..
... . ..
Aromatics
e 33.7 34.9 42.3 38.3
1.5578 1.5609 1.6040 1.8848
therefore unexpected. This line of reasoning could further lead one t o propose that much higher viscorities of the residua arc the explanation for the diffcrenre. However, the aromatics analyses in Table I V are b a d on weight and i t is therefore quite likely t h a t the aromatics from the high molecular weight rrsidua contain, on the average, smaller condensed ring species (with correspondingly longer alkyl side chains) than do the aromaticq in the gas oil feed streams. Chromatographic analysis of sample 114 has shown ( 1 3 ) that very little of the aromatirs from this residuum appears t o be above the range of, sav, alkvlated naphthalenes, with further evidence that mononuclear aromatics with a high degree of long chain alkyl substituents are aleo preqent. Incontrast, an extensivechromatographicanalysis of the 700" F. aromatics from a virgin gas oil feed similar t o those of Tahlr TI'
+
August 1952
I N D U S T R I A L A N D E N (Z I N E E R I N G C H E M I S T R Y
has shown t h a t in addition t o about 74 (weight) % of benzenes and naphthalenes, there were also present about 15% of phenanthrenes, about 6% of four-ring aromatics indicated t o be pyrenes, chrysenes, and benephenanthrenes, about 2% of five- and higherring aromatics, and about 2y0 of unknowns. These results suggest a chemical explanation, or possibly a combination of chemical end viscosity effects, for the relative tumor potency values of the virgin gas oils and the crude residua. However, this point cannot be resolved with the present data, especially since viscosity itself is a function of chemical composition.
T4BLE
v.
Samplo No. 429-48 430-48 431-48 432-48 480-53
TUMORPOTENCIES O F SOLUTIONS CARCINOGENS
OF
VARIOUS
Description
Tumor Potency
in dodecyl0 1% of 1,2,5,6-dibenzanthracene benzenea 0 . 1 % of 9,10-dimethy1-1,2-benzanthracene in dodecylbenzene 0 17 of 3,4-benzpyrene in dodeoylbenzene 0 1 4 of 20-methylcholanthrene in dodecylbenzene 0 15% of 20-methvlcholanthrene in dodeovl-
222 59
68
106
Description of this solvent given in text. The unexpectedly lower tumor potency value for sample 480 as compared with sample 432 may be due t o error.
The general range of tumor potency values shown thus far for high boiling virgin streams including various slack waxes, lubricating oil stocks, virgin gas oils, and crude residua is from zero up t o low or moderate values, that is, from zero u p t o occasionally about 30. Some reference data for comparison are given in Table V, which contains tumoi potency values for solutions of several of the most potent known polynuclear aromatic hydrocarbons. Customary concentrations were employed and the same kind of mice and testing conditions were used as in t h e testing of the oil samples. As is well known, the tumor response depends on the solvent. The preferred solvent in most of these tests was a stock sample of dodecylbenzene. This sample was an isom'eric mixture of dodecylbenzenes, boiling from 520' t o 568" F., prepared by alkylation of benzene with a commercial polypropylene fraction in the presence of aluminum chloride, having the properties: diO, 0.8715; ng, 1.4897; bromine number, 0.4; and 90% of sample boiling from 540" to 551" F. This was chosen because its chemical nature, its high boiling range, and its weight ratio of alkyl radical t o aromatic ring content cbl-respond t o these properties of high boiling petroleum fractions much more closely than do the customary solvents, benzene or acetone. Also, it was found t o be a biologically satisfactory solvent under the animal test conditions employed in these studies (11). CATALYTIC AND STEAM-CRACKED STOCKS
Preliminary tests showed t h a t high boiling catalytically cracked cycle stocks, slurry oils, clarified oils, and steam-cracked tars have high tumor potency values. Tumor potencies of 55 and 63, shown in Table VI for a catalytically cracked cycle stock and a steam-cracked tar, are typical of these materials in the boiling ranges indicated. The two stocks of Table VI were examined for the influence of boiling range on tumor potency. Table VI1 shows t h a t fractions of the cycle stock boiling below 700" F. have zero tumor potency values. Fractions boiling above 700" F. are quite potent and show roughly constant tumor potency values over a wide boiling range. The aromatics analyses in Table VI1 show t h a t the observed dependence of t h e tumor potency values on boiling range is not due t o aromatics enrichment arising from distillation, since t h e aromatics contents of the fractions remained quite constant over most of the range.
1821
However, the refractive indices of t h e aromatics from the fractions increased with boiling range, and there is some evidence of a possible maximum in refractive index of the aromatics. Since a given carcinogenic component of polynuclear aromatic type will undergo some rectification when distilled over the range investigated (4),the results of Table VI1 suggest t h a t the tumorproducing activity of a high boiling catalytic cycle stock arises from a multiplicity of carcinogenic components having a wide range in boiling points. This suggestion receives further support from tests made on chemical fractions, briefly reported elsewhere (6). The steam-cracked t a r of Table VI gave distillation cuts showing a temperature variation of tumor potency values t h a t was qualitatively similar t o that from t h e cycle stock. The results are given in Table VIII. The unexpected appearance of low hut nonvanishing tumor potency values for the cuts boiling below 700" F. is attributed t o faulty distillation or entrainment or to the distorted vapor pressure phenomena often encountered in multicomponent hydrocarbon systems, because the redistillation data for the cuts in question showed substantiaf percentagw of material boiling outside of the nominal range, as is seen in t h e table. These results indicate t h a t t h e low potencies for the t a r fractions boiling below 700" F. may be due t o the inclusion of some 700 " F.+ material in these cuts, but i t remains to be established whether the same potency threshold holds for the steamcracked tars as for catalytically cracked cycle stocks. Similar anomalies have been observed on a few occasions with distillation cuts from cycle stocks and clarified oils boiling below 700"F. I n each instance, these apparent exceptions have been resolved by careful redistillation. Again, t h e trend in tumor potency values with boiling range cannot be explained through differences in the aromatics contents of the t a r cuts, although t h e refractive indices of the aromatics increase with the boiling ranges of the whole cuts.
TABLEVI.
INSPECTION DATAFOR HIGHBOILINGCRACKED SmcgS
Description Sample No. Tumor potency Gravity, O A P I Viscositv. SSU a t 100° F. Av. moE wt. Val 7' a t 700°. F. vel: aromatios Vacuum distillation, O F. (atm.) a t
%
+
Catalytic Cycle Stock 191 55 26.2 68.9 282 48 32 643 697 830
Steam-Cracked Tar 109 63 -0.8 5400 280 50 95 (approx.) 524 706 945
Table VIII shows t h a t the steam-cracked t a r fractions boiling in the broad vicinity of 800' or 900" F. may have maximum tumor potency values. The evidence is not entirely conclusive because the highest boiling fractions and bottoms from the cracked oils and tars could not be directly tested on mice, owing to their extremely high viscosities. However, Table IX shows t h a t certain concentrated dispersions and solutions of t h e 1050" F.+ bottoms (sample 399) from the steam-cracked tar 109 gave either very low or zero tumor potency values, and i t is inferred from these results that if these bottoms materials could have been tested without dilution, t h e resulting tumor potency valuea would probably have been less than the highest values recorded in Tables VI1 and VIII. These indications of maximum tumor potrncy values at intermediate boiling ranges receive some independent support from data in the literature. Figure 1 shows how t h e tumorproducing activities of a large number of pure aromatic hydrocarbons depend on molecular weight and ring number. T h e activities were estimated from data compiled by Hartwell (6)
I N D U S T R I A L A N D E N G I N E E R I N G CHEMISTRY
1822
Figure 1. Correlation of Tumor-Producing Activity with Molecular Weight and Ring Number for Various Aromatic Hydrocarbons 'rumor activities were estimated from Hastwell's data (6)
and must be regarded as qualitative. (Hartwell's ditta represent a large variety of test conditions, different numbers, strains, and species of animals, various concentrations of aromatics, and different sites and frequencies of applicatioii. Tumorproducing activities estimated from such data therefore cannot be compared numerically, but do serve t o show in a broad sense whether t h e materials are carcinogenic or not.) The plot shows that the potent Pithstances lie within a molecular
TABLE VTI.
TUMOR POTENCIES OF DISTILLATION C'tiTS HEAVYCATALYTIC CYCLESTOCK 191
OF
(15 plates: 2:l reflux ratio: 10 mm. of mercury pressure) Sample No. 404-46 405-45 406-45 407-45 408-45 409-45 410-45 411-45 412-45 413-45 414-45
Boiling Range,
F. (atm.)" GOO-650 650-700 700-750 750-800 800-900 900-930 93 0 -9 50 960-965 965-980 980-1010 lOlO+ bottoms
Yield, Vol.
%,
24.1 29.1 21.4 I1 1 1.9 2.1 2.1 2.1 2.1 I .5
2.5
Tumor Potency 0 0 48 34 46 62
53 -5 8 G5 48 , .
Viscosity, 8SU a t 100' F. Wt. % 48.5 32 0 54.1 33 2 67.6 30 F 93.5 30 6 148 31 4 150 32 0 158 33 0 34 4 180 220 37 8 42 8 308
b t . .
1 .G140
1.6262
1 ,6342
1.6370 1.6797 1.6665 1.6664 1.663'1 1 ,6660 1 6672
...
Cuts having atmos heric boiling points above 800" F. ivere takori si e streams from t h e &stillation column. Too viecous for direct testing. a
:LS
db
weight range of 230 t o 320, wliich embraces compounds having from four t o six condensed aromatic rings. Figure 2 is a similar plot of tumor p0tenc.v values against average mo1ecul:Lr weight for a large number of high boiling petroleum refinery samples tested in this work. These included a large variety of product types, fractions, and hlendp. Also, their aromatics contents, as well as their widely differing molecular weight distribut,ions, varied considerably. These differences may explain why t,he potency boundaries on the molecular weight scale are not as sharply defined as in Figure 1. The correspondence is nevert,heless surprisingly good. A similar plot was obtained foi the pet#roleumsamples when the midboiling point was employed instead of average molecular weight. It appears from these plots that for oils containing complex mixtures of condensed ring aromat,ics, the association of ma,ximum tumor potency with intermediate zones of average molecular weight or inidboiling point, arises from the likelihood that the latter are, :ii. least, roughly, measures of average ring number. Examination of aromatic extracts and the corresponding raffinates from high boiling catalytically cracked products has established, as in the case of slack waxes, t h a t the carcinogenicity is associated with the aromatic components. Some typical data,
Vol. 44, No. 8
are given in Table X. These show t h a t when the aroinativs separation is clean-cut (as through suitable silica-gel percolation) one ends u p with a n aromatics extract having a tumor potency value higher than the parent stock and a corresponding nonaromatic raffinate having n zero or insignificant tumor potency value. The inordinately high tumor potency value recorded in Tablr X for the pressed wax 102 cannot be explained with the presenl data. Tests of many other wax samples always yielded Iom tumor potency values. Also, Table X shows t h a t the tumoi potency value of this wax sample is not in accord with the tumor potency values of the corresponding aromatic8 extract arid raf3nate, nor does its relatively high boiling range agree with those of the other slack waxes examined (see Table I). Results similar to those in Table X were obtained with othei high boiling cracked products. The tabular data also show that raffinates (such as sample 105-15) which are not complete13 nonaromatic (as from sulfur dioxide extraction) can retain a tumor potency corresponding t o their aromatics content. I11 testing distillation cuts, extracts, or other fractions, controlb were included, whenever sample sizes were adeqiiate, of reblends of the fractions t o reproduce the original stock. This was done t o ensure t h a t the properties of the fractions were not altered h v the separation process. T H E R M A L CRACKING
CO\YEYTIONAL
Table XI contains data for a highly aromatic, tar fioni the theiinal reforming of heavy virgin naphtha. A eamplc of the ferd stock was not available for testing. The tar sample as found to have a moderately low potency comparable t o that of a virgin gas oil feed t o catalytic cracking, shown in the table for comparison. However, since the boiling range of a heavy virgin napht1i:i lies far below 700" F , tlicw very limited results 4io8- that carcinogens are s>n tliesiwed in conventional thermal rl a c k i n ~ Howrver, tar4 from thermal cracking of virgin naphtha and from reduced crudes h a t e much loner tumor potency vnlurs than do tars from c.atalyticall> cracked gaR oil%and from *team-cracked parafinir distillates DISPOSAL OF HIGH BOI LIM;
c 2 i r r A L n I ( : CYCIA
wocx
Animal tests obtained to date show without eaceptioii that catalytic cycle stock and related products (slurry oil, clarified oil, catalytically cracked t a r ) boiling above 700 O F. have high tumoi potency values. The potential hazard of such streams therefore poses a major problem t o thi. refinei, because substantial volumes of such cycle stocks are produced. .41so, much of 1his output is blended into bunker fuels or similar heavy fuel blends IT hich may be used under conditions requiring a low tolerable limit of hazard. Various relatively elegant arid expensive physical and rhrmical prow'sirw can be proposed for the p ~ n p o s eof reduc'iii~the (-:wino-
9-,
100-
,'o
0 L--
IW
.
o
oo
c
33im
_
150
200
\
250
.
300 AVERaGE
M 0 _
350
400
MOLEOULAR
WEGHT
450
0
-
_ -__-A
500
550
600
Figure 2. Correlation of Tumor Potency Values with Average Molecular Weight for Various High R o i l i n g Plant Samples, Fractions, and Blends
-
August 1952
INDUSTRIAL AND ENGINEERING CHEMISTRY
1823
POTENCIES OF DISTILLATION CUTSOF STEAM-CRACKED TAR109 TABLE VIII. TUMOR Sample Boiling Range, Yield, Tumor Gravity, NO. F. (atm.) Vol. % Potency 'API 393-45 Below 600 27 4 25 394-45 600-650 15 4 25 7'2 650-700 15 2 28 4 1 395-45 396-45 700-750 6 2 73 1 4 750-800 5 4 100 397-45 800-1050' 6 3 85 398-45 1050+ bottoms 24 1 . f 399 a The aromatics determination method includes a n evaporation ,step which results in falsely low values for samples containing appreciable amounts of aromatic material boiling below about 600' F. b At 98% over.
Viscosity,
SSU at
.. .. ..
Redistilled Cut Boiling range, 0 F.
looo F.
Zti3
36 9 53 3 66 4 156 949
20 15 15
so
*
650-672 700-720' A4bove750d
....
... ...
Aromatics Wt. % n 56 4a 1 6025 1 6282 81 4a 1 6317 86 8 1 6556 91 8 1 6489 95.0 1 6911 98 2
..
At 97% over, Also 27 weight % below 700° F, e Side-stream out. I Too viscous for direct testing.
TABLEIX. TUMOR POTENCIES OF THE 1050' F.+ BOTTOMS FROV STEAM-CRACKBD TAR109 (SAMPLE399) Sample NO. 400-45
Turnor
Description
Potency
50 wt. % of finely groundn sarnpp 399 dispersed in U.S.P. petrolatum
0 50 wt. % of sample 399 p o l v e d in 50 wt. % of C.P. pyridine 20 402-45 10 wt. % of sample 399bdissolved in 25 90 wt. yo of C.P. toluene 100-mesh and finer. Mouse tests on t h e petrolatum, pyridine, and toluene controls were negative 401-46
*
genicity of these streams or blends. However, the simplest alternatives available a t this time for the disposal of such stocks, which are consistent with economic and other limitations, are: 1. Further processing (recycle catalytic cracking or thermal cracking) to reduce production of 700" F.+ catalytic cycle stock. 2. Blending with noncarcinogenic refinery streams to what is considered a safe level. 3. Use of the 700" F.+ catalytic cycle stock under conditions that minimize human contact. This last alternative is a hygienic rather than a technical solution to the problem and will not be further considered here, especially since a hygiene control program has been published elsewhero (7'). RECYCLE CATALYTIC CRACKING
Where spare catalytic cracking capacity is available, the volume of 700" F. clarified oil can be reduced or cracked t o extinction by recycle catalytic cracking. Since cracking t o extinction is not generally economical, the present practice in some cases is to reduce the quantity of clarified oil by recycling or combining with thermal cracking t o such a n extent t h a t the remainder can be disposed of by blending into bunker fuel without exceeding the
+
TABLE x. TUMOR POTENCIES Sample
No.
101 102-15 103-15 104-15 112-15 158-21 136-13 105-15 a
AND INSPECTION
f present informatiorr,
' less than 10 volume % ve 700" F. can be considered safe and require no special precautions. Experiments leading to this restriction are described elsewhere ( I I ) . ] Mouse tests and chemical analyses of clarified oil were carried out t o learn how recycle catalytic cracking influences the carcinogenicity of the oil. Table X I 1 gives analytical and production data for clarified oil produced at, 0 and 12.7% recycle in a commercial cracking unit. With regard t o the condensed ring aromatics in the 700' F.+ fraction-components of which are undoubtedly the main if not the entire cause of the carcinogenicity of clarified oil-it is seen that the catalytic cracking step increases the concentration of these by nearly tenfold their concentration in the virgin feed. On the other hand, their concentrations are nearly the same in the run containing no recycled oil and t h a t in which 12.7% of the clarified oil was recycled with the fresh feed. These results indicate t h a t recycling of clarified oil does not increase seriously the concentration of those aromatics which include the carcinogenic types. The data also show that, based on fresh feed, the catalytic cracking ate4 gave somewhat more aromatics of four or more condensed rings (of altered structure) than were present in the fresh feed. However, recycling t o t h e extent of 12.7% on fresh feed destroyed nearly half of these polynuclear aromatics. It is therefore evident that under constant cracking conditions recyclingof 700" F.+ material t o thecatalytic cracking step destroys some of the polynuclear aromatics. I n addition to recycling, other variables, such as feed stock coniposition and cracking conditions, also probably have some influence on the yield of polynuclear aromatics., However, insufficient data are now available t o define t h e effects of these,variables. Table XI11 gives production data for commercial cracking runs carried out a t two refineries under various recycle conditions.
it is believed that
DATAFOR VARIOUB AROMATIC EXTRACTS AND CORRBSlWNDING RAFFINATES Tuinor Gcavity, Potency API 55 24.6 36 83 (?) 67 4.6 0 42.7 77 4.8 12 40.8 -10.8 59 30.3 45
Description Heavy catalytic cycle gas oil Wax pressed from 101 at 25O F. Aromatics from SiOe-gel extraction of wax 102 R a 5 n a t e from SiOr-gel extraction of w&x 102 Aromatics from StOl-gel extraction of oil 101 R a 5 n a t e from SiO2-gel extraction of oil 101 Aromatics from SO1 extraction of dewaxed oil 101 R a 5 n a t e from SO2 extraotion of dewaxed oil 101
Viscosity, Average Vol. SSU at Mol. 100' F. Wt. 7 0 3 285 67.1 40 294 35.3a 100 252 ... 313 34.76 .. 252 .. 214 306 53.4 ,. 247 255 50 314 67.6 40
Vol.
%:+ %matics 32 15 I00
..
0 100
0
95
Vacuum
F., a t Vol. % '
5% 644 890
..
50% 683 1100
95% 800
700 687
870 826
..
644 649
20
..
.. ..
.,
At 210° F.
TABLE
Sample NO. 361-42 110-33
XI. TUMOR POTENCIES Description
Vifgin feed t o catalytic cracking Thermal reformer t a r
AND INSPECTION
DATAFOR
Tumor Gravit Potency OAPIY' 33
28.4
32
8.2
Viscosity, SSU at 1OO'F 62.9 315
4
THERMAL REFORMER TARAND Average Vol. Mol % at Wt.' 700' F.+
A
VIRGIN GAS OIL FEED
Vacuum Distillation, e F., at Vol. % 5% 50% 9,577
274
45
496
676
258
45
470
658
923
Aromatics Wt. % nLo 33 7
1.6578
84.0
1 6109
INDUSTRIAL AND ENGINEERING CHEMISTRY
1824
T ~ R L EX I . ANALYTICAL AND PRODUCTION DATAFOR CLARIFIED OIL FROM A COMMERCIAL CRACKING UNIT Gas Oil Feed Yield of clarified oil vol % Yield of 7OO0,F.+ ilarified oil, vol. ToCal aromatics above 700' F.. mt. @% I n 700" F.+ fraction ' 'Based on fresh feed Total condensed ring aromatios with four or more rings, wt. % I n 700" F. fraction Based on fresh feed
+
*
Clarified Oil ---from--0% 12 7% recycle recycle test test 12.6 6.7 5.7 11.1
35
44
49
21
2.5
1
~~
2.5
20.3
1.5
2.2
22.2 1.3
a Determined by silica-gel absorption. Determined by repercolating the SiOz-gel aromatics through a 10-section alumina column, using n-heptane &a t h e developer. T h e aromatics in the individual sections were then desorbed with acetone the solvents were removed under vacuum a t 200" F., a n d the aromatic tfpes of the resulting fractions were determined from the ultraviolet spectra and from additional d a t a , including carbon a n d hydrogen analysis, molecular weight, and infrared spectra (8).
*
T ~ B L E XIII. P R O D U C T I O N DATA FROM REFIKERY&CYCLE RUNSAND C O R R E S P O N D I N G TUMOR P O T E N C I E S O F ('LARIF IEI) OIL FUELBLENDS Refinerv ___ -~.4 11 18 Recycle in total feed vol. yo , Over-all 430" F. cdnversion, In vol. % ' 6 1 . 2 61 4 of fresh feed Yield of 700° F. clarified oil, in vol. % 2 1 2 5 of fresh feed Tumor potency of synthetic fuel blendn containing 20 vol. % of 700" F.+ 40 35 clarified oil
+
+
0
Refinerv I1 ___10 40
54 146
36
~
56 7
60 0
9.7
4 6
43
48
Blended b y diluting the clarified oil with a base stock (300 SSF a t 122" F.) of West Texas arude residuum a n d U.S.P. white oil. Table, I1 and I V show the latter two materials (samples 108 a n d 114) t o be completely noncarcinogenic.
Vol. 44, No. 8
tion in the nonaromatics, this reduction is found from the data t o be inadequate t o account for all of the lower boiling fraction. produced. That is, the latter products arise, at least in part, from the cracking of the aromatics in the catalytic feed, even though these aromatics are commonly regarded a s being very "refractory.") Likewise, the concentration of aromatics having four or more condensed lings is greatly increased by the thermal cracking; unreported data show that roughly 80 ncight % of the resulting fraction boiling above 700' F. consists of such aromatics. In addition, the tabular and other aromatics analyDes show that on a feed basis thermal and steam cracking gave high boiling products containing from 50 to 70 weight %more four and higher condensed ring aromatics than were present in the feeds. One might expect on these grounds that t h e tars from the thermal and steam cracking of clarified oil should be more carcinogenic than the corresponding feed stocks. However, the data show that the average tumor potency values of the feed and t a r from the steam cracking run do not differ significantly and that the tumor potency value of the tar blend from lahoratory thermal cracking is not much higher than that of its corresponding feed blend, if indeed the differenceis significant a t all. This indicate8 that these cracking processes have produced in the higher aromatics biologically important structural changes which are not revealed by the changes per se in higher aromatics contentc It can be concluded that thermal and steam cracking of claiified oil has the beneficial effect of converting a substantial part of the oil into innocuous laxer boiling products without a corresponding or appreciable enrichment in the tumor potency of the refiidual tars. Table XV gives some changes in composition which occui in various high boiling aromatic types as the result of various cracking processes. The separations and analyses were rarried as briefly described in a footnote of Table XIT. (Because of spectral and other evidence that appreciable overlapping can occur in the chromatographic fractions from the analytical method, the naming of the various ring types of Table XV is only roughly descriptive and the corresponding rinalyticnl oyit
Also included are the tumor potency values of synthetic heavy fuel blends containing the corresponding clarified oils. The indicated blending agents were employed t o eliminate the possibility t h a t any of the observed tumor potency might be due to the blending stock. The production data shown in the table confirm that the greater the recycle ratio, the more complete is the destruction of the 700"F.+ clarified oil and polynuclear aromatics. The potency data for the heavy fuel blends are less conclusive. The blends from the refinery A runs suggest a slight, probably insignificant, decrease in the tumor potency of the clarified oil with increased ratio of clarified oil recycled t o catalytic cracking, while those from the refinery B runs suggest the opposite effect. Unfortunately, the lSyo recycle run made a t refinery A employed milder cracking conditions than did the 11% recycle run. The refinery B results, which are directly comparable, indicate t h a t even though t h e tumor potency of the clarified oil fraction may increase somewhat with increased recycle ratio, this effect is more than compensated by the decreased production of the fraction.
XI\'. THERMAL AND STE.4M CR.4CKINC CATALYTICALLY CRACKED CLARIFIED OILS
TABLE
Pressure lb./square in. Coil outiet temperature, 0 F. Operation Fresh feed in total feed, vol. %
It has been demonstrated t h a t thermal cracking of catalytically cracked clarified oil boiling above 700" F. will greatly reduce the volume of this material by conversion to lower boiling fractions. Table XIV shows data for a 700" F.f fraction of a clarified oil that was thermally cracked in a laboratory run and also for a laboratory steam-cracking run made with an 88 volume % overhead from a different sample of clarified oil. The feed stock and t a r Kere analyzed for total aromatics boiling above 700" F. and fer polynuclear aromatics having four or more rings. The data show that thermal and steam cracking greatly increase the concentration of aromatics, especially in the fraction boiling above 700" F. I n fact, this fraction is shown t o contain only from 0 t o 16 weight yo of nonaromatirs after the thermal cracking operations. (Although the thermal cracking reeults in a large reduc-
Laboratory Laboratory Steam Cracking Thermal Cracking of 88% Overof Clarified head from Oil Boilin Clarified Oil above 700°%. 25 750 1040 1250 Coil only Downflow soaker 100 100
Yields Gas, wt. 70 CAt o 430' F., vol. 430' t o 650' F vo?% 6.50' t o 700' F:: vol. % T a r , vol. %
THERMAL CRACKING OF CLARIFIED OIL
O>
27.0 33.0
9.8 32.7 6.5 2.2
141.7
48.0
430' F.+ _____ Feed Tar
- 700' F.+ __ Feed Tar
63, 66
36. 3 8 b 45, 5 3 b
65
4 4 , 70 165
iiih
iiib
45
50
100
100
Analyses of feed and tar Total aromatics above 700° F. On sample, wt. 7*, On 700° F.+ fraction, wt. 7; On feed, wt. yo
14.8 36.0 14.8
38.7 84.0 18.3
69.0 59.0
86 100 38
Polynuclear aromatics, four 01 more rings On samnle, wt. yo On 7003 F.+ fraction, wt. % On feed, wt. %
10.1 25.2 10.1
31.8 74.1 15 0
..
, . .
59.0
..
Repeated mouse tests. For a blend of 25 volume 7 of feed or t a r and 75 volume c' of a base stock (300 SSF a t 1220aF.) o f West Texas crude residuum 2 n d U.R.P. white oil (as in footnote of Table XIII). a
INDUSTRIAL AND ENGINEERING CHEMISTRY
August 1952
TABLEXv.
CHANGES IN
AROMATICCOMPOSITION AS RESULT OF VARIOUSCRACKING PROCESSES
Aromatic Ring Type Benzenes Naphthalenes Anthracenes Phenanthrenes Pvrenes Binaanthracenea Chrysenea and benzphenanthrenev Five rings Unknown (to 1020° F. cut point) Unknown (above 1020' F. cut point) 450° t o 930° F. cut from West Texas crude. 88 volume % overhead from a olarified oil. Low, due to benaanthraoenes in pyrenea fraction. High, due t o benzanthraeenes.
+
a
r'
I
Weight Per Cent of Ring Type in Aromatics Fraction Boiling above 700° F. Refinery Catalytic Cracking of Heavy Gas Oil Laborator s t e a m Cracking Refiner Thermal Cracking Clarified oil Clarified of d a r i f i e d Oil orclarified Oil (12.77 Fresh (0% Feedb 43Q0 F.+ T a r Feed 43Q0 F.+ T a r feed" recycle) recycle? 60 6 6 1 0 0.4 0 14 5 3 0.2 0.2 0.6 0.1 0.6 1 2 1 1 1 0.7 16 22 17 30 16 81 14 3 21 22 49 40 22 26 0 4 1 2.J" 22 2; 15 3 24 4 2 9 13 2 4 6 2 0 6 0 0 0 0 0 9 8 0 20.9 10 29
135:
1
I
T
I
-I
50-
i
6 E 400 c
30-
i: 20 -
10-
'L
-0
0
IO
v ,VOL. PER CENT
ceir
1825
GAS OIL IN BLEND
Figure 3. Dependence of Tumor Potency Value on Composition of Blends of Heavy Catalytic Cycle Gas Oil (Sample 101) i n West Texas Crude Residuum (Sample 114)
Vertical dotted lines show the 19/20 coufideuce limits of the tumor potency values from single mouse t e s t 9
values are therefore only indicative. Overlapping in the fractions appears t o vary between different samples and i t is not yet possible to assess the probable error in the ring-type descriptions and analytical data of the table.) It is seen that in catalytic cracking of a heavy virgin gas oil the high boiling benzenes and naphthalenes, which comprise the major portion of the aromatic components, are greatly reduced in concentration and ring condensation occurs t o give a large increase in the concentration of aromatics having from three t o five Condensed rings. Appreciable conversion t o more complex ring structures (five and more rings and very high boiling unknowns) also results from the cracking process. (The data are not sufficiently exact and extensive t o define the effect of recycle ratio on the conversions of the various ring types.) The data for the steam cracking and thermal cracking of catalytic clarified oil feeds have a common pattern. Very little of the high boiling benzenes, naphthalenes, and anthracenes is present in these feeds and little if any change in these components results from the cracking processes. The largest effect appears t o be the decrease in phenanthrenes, with only modest changes occurring in four and higher condensed ring types. The conversion of phenanthrenes appears to account for a considerable portion of the very high boiling unknowns present in the tars, although the mechanism of the conversions and redistributions in aromatic types occurring during cracking is doubtless far from being so
228:
1
simple. These analyses appear to be consistent with the previous observation, made in connection with Table XIV, that no appreciable enhancement was found in the tumor potency values of the residual tars from steam cracking and thermal cracking of clarified oils; if such an enhancement had occurred, one would have expected it to be accompanied by a corresponding increase in the concentrations of aromatics containing from four t o six condensed rings. The moderate enhancement in the tumor potency value of the t a r blend from laboratory thermal cracking, shown in Table XIV, is paralleled by the increase in pyrenes shown in Table XV. However, it is doubtful whether either the biological or analytical data are precise enough for this parallel t o be considered significant. BLENDING
Disposal of high boiling catalytic cycle stocks and tars by blending with noncarcinogenic refinery streams t o a safe level is a simple alternative that finds its most important application, from a refinery viewpoint, in the blending of clarified oils into bunker fuel oils. For this reason, the effectivenessof blending in reduc-
$01 40
1
I
I
0
i
i
IO
20
30
40
50
80
70
80
90
100
VOL. PER C E N T REFORMER TAR IN BLEND
Figure 4. Dependence of Potency Value on Composition of Blends of Thermal Reformer Tar (Sample 110) i n West Texas Crude Residuum (Sample 114) Vertiaal solid lines show range of tumor potency values from repeated mouse teats
ing carcinogenicity was tested in blends representative of bunker fuels. Table XVI gives the tumor potency values and inspection data for the components of these blends. The blends were composed of varying ratios of a heavy catalytic cycle gas oil (sample 101) in a West Texas crude residuum (sample 114), a thermal reformer tar (sample 110) in the same residuum, and a clarified oil (sample 173) in a Gulf Coast crude residuum (sample 175). Figures 3, 4, and 5 show how the tumor potency value varies with the volume percentage of the carcinogenic component pres-
INDUSTRIAL AND ENGINEERING CHEMISTRY
1826 TABLE
Sample
XVI
Tu>foR P O T E N C I E S
No.
Description
101 110 173 114 175
Keavy catalytic cycle gas oil Thermal reformer tar Clarified oil Residuum from West Texas crude Residuum from Gulf Coast crude
IPoc----
-
,
iN11 I N S P E C T I O N
24.6 8.2 18.6 10.8 21.4
55 32 65 0 0
,
,
..-
Vol. 44, No. 8
DATAOF C O M P O N EUSED ~ T ~I N BLENDING STUDIEL
67.1 316 110 140 000 3:OOO
285 258 250 793 400
40 45 75
100
82
644 470 604 879 574
683 658 760
880
800 895
.. ..
37.0 84.0 42.4 42.3 38.3
1.6262 1.6109 1.6750 1.6040 1.5843
._
'7
i
Bo-
Figure 5. Dependence of Tumor Potency Value on Composition of Blends of Clarified Oil (Sample 173) in Gulf Coast Crude Residuum (Sample 173) Significance of vertical solid and dotted lines as described in Figuree 3 and 4
ent in the blend. I n all cases, the tumor potency value increases in direct proportion t o the volume percentage of the carcinogenic component present, up t o about 60%. Beyond this point, it cannot be decided from the available data whether the curves really pass through a maximum, even though the plotted points suggest this possibility. The figures show that, within the limits of error of the data, the tumor potency curves can be considered l o break in the vicinity of 60010, with little or no further change occurring in tumor potency value as the ratio of the carcinogenic component is further increased. That is, the tumor potency values may show a saturation effect. Also, they are higher than would be computed on the basis of additivity. For blends similar t o those studied, i t is seen that approximate tumor potency values can be predicted from the values of the two components, if a 60% break point is used in constructing the two linear portions of the curve. On first thought, the occurrence of a break point might be considered explainable on the basis of viscosity. T h a t is, it might be speculated t h a t a threshold viscosity exists, beyond which viscosity interferes with the penetration of the material into the skin, with a resulting decrease in potency corresponding t o the initial portions of the curves shown in the figures. However, calculation shows t h a t the break points are not isoviscous; the viscosities (SUU a t 100" F.) of the three blends of Figures 3,4,and 5 at t h e break point are found t o be 526, 514, and 2700, respectively. More data are needed to test this possibility. When the data of Figures 3, 4,and 5 are replotted on the basie of the 700" F.f material in the potent component, rather than the whole material, the curves are, of course, merely shifted on t h e horizontal axis but retain the general features shown. It is found t h a t when not more than 10 volume % of 700" F.+ catalytic cycle gas oil or clarified oil is present, the tumor potency values of the blends with crude residua are less than 20 and therefore have borderline significance. The thermal reformer t a r blends appear t o have a somewhat higher tolerance limit, as can be judged from the relative flatness of the curve of Figure 4 in
tho low range. The pobitive deviations from simple additivity shown by the tumor potency values of the blends cannot be explained with the present data. SUMMARY
An extensive study is reported of modern refinery streams and blends found to be carcinogenic in certain tests on lower animals, and therefore assumed t o be potentially hazardous t o man. ;\lain emphasis is directed toward fractions and blends containing high boiling catalytically cracked streams and the relationship of their tumor-producing capacity t o refinery processing, but results are included for virgin stocks including slack \Y,~XPS, lubricating oil stocks, virgin gas oils, and crude residua. High boiling catalytically cracked streams (heavy cycle stocks, durxy, and clarified oils) had high potencies. Fractions boiling above 700" F. are potent, while those boiling below this temperature are innocuous. Similar results are shown for fractions of steam-cracked tars. The potency is associated entirely with the aromatic components. The potency of distillation fractions varies with boiling range, with inconclusive evidence of an intermediate range of maximum potency for the tar fractions. Lower potencies are inferred for very high boiling bottoms of catalytically cracked stocks; these had to be tested as dispersions and solutions. Virgin gas oils had moderate potencies commensurate with those expected if the parent crudee possessed a low order of carcinogenicity associated with the higher boiling fractions. Crude residua were completely devoid of potency. The difference in potency between virgin gas oils and crude residua mag be due t o chemical rather than viscosity differences, Tars from thermal cracking of virgin naphtha and from reduced crudes are much less potent than tars from catalytically cracked gas oils and from steam-cracked paraffinic distillates. Petroleum slack waxes had very low or marginal potencies. There were no significant differences in potency for a series of slack waxes pressed from a given crude or for slack waxes from t h e different crudes examined. The tumor-producing components of slack wayes are associated with the aromatic fractions of the waxes. Highly refined waxes and related products (petrolatum and white oils) which have undergone removal of aromatic components are devoid of potency in the animal tests. Most of the lubricating oil stocks and their phenol extracts were completely innocuous; a few samples had low potencies. Conventional acid treating appears t o have no decisive effect on potency. Two of the phenol extracts had zero potencies and one had a low potency. These results are attributed t o the corresponding potencies of the parent stocks. The amount and refractive index of the aromatics fraction from silica-gel separation of a lubricating oil stock appear to be without value for predicting the potency of the stock. This result is explained by the nature of the polvnuclear aromatics present in virgin stocks. Safe disposal of catalytic cycle stocks and clarified oils can be accomplished economically by recycle catalytic cracking or thermal cracking t o reduce production of catalytic products boiling above 700' F. or by blending t o a safe tolerance level with noncarcinogenic refinery streams. D a t a are given on the effect of recycle catalytic cracking on the production of high boiling frac-
August 1952
IN D U S T R I A 2 A N D E N G I N E E R I N G C H E M I S T R Y
tions and on the changes in aromatics composition accompanying the partial destruction of polynuclear aromatics by the cracking process. Similar data are given for the thermal cracking and steam cracking of clarified oils. Data are also given t o show how the potency varics with blending ratio for blends of bunker fuel type containing high boiling cracked oils or thermal reformer tars and crude residua. Potencies of such blends are not simply additive; they increase in direct proportion t o the percentage content of the cracked component up t o about 6070, and then remain uncahanged a t higher perrentages, within the limits of error ACKNOWLEDGMENT
The authors w e pleased t o acknowledge the large amount of experimental work carried out by B. F. Dudenbostel, Jr., L. T. Eby, R. L. Mathiasen, G. G. Wanless, and other laboratory associates. Special thanks art) due to H. G. M. Fischer for his enthusiastic and able guidance and advice in t h e program. The extensive animal test data used in this work were obtained and supplied by W. E. Smith and D. Sunderland of the New York University Hellevue Medical Center, and by K. Sugiura of the Sloan-Kettering Institute. R. E. Eckardt of these laboratories, and W. C . Hueper of the United States Public Health Service, contributed much valuable discussion on the hygienic and medical aspects of the problem. c. L. Brown, R. M. Shepardson, and M. W. Swaney contributed valuable support and technical disrussion. Finally, thanks are due t o the managements of the Standard Oil Co. (New Jersey), the Esso Standard Oil Co., and
182?
the Standard Oil Development Co., for their generous financial support of t h e project and their encouragement t o publish t h e results, and t o the managements of affiliated refineries for their cooperation in the experimental plant rum. LII‘ERA’I‘URE CITED
Badger, G. M.,Brit. J. Cancel., 2, 309 ( 1 9 4 8 ) . ( 2 ) Hlanding, F. H., King, W. H., Jr., Priestley, W., Jr., and Rehner, John, Jr., Arch. I d . Hug. Occup. Med., 4, 335 (1951). (3) Charlet, E. M., Lanneau, K. P., and Johnson, I?. D., preprint, (1)
p. 105,
Petroleum Division Symposium, 1 1 9 t h Meeting
AM. CHEM.SOC., Cleveland, April 9, 1952. (4) Eby, L. T., Wanless, G . G., and Rehner, John, Jr., IND.ENG.
CHEN.,43, 9 5 4 (1951). ( 5 ) Fischer, H. G. M., Priestley, W., Jr., Eby, L. T., Wanless, G. G., and Rehner, John, .JI... Arch. Ind. Hyg. Occup. Med., 4, 315 (1951). (6) Hartwell, J. L., IT. 8 . Public Health Service, P u b . Health B u l l . 149 (1951).
(7) Holt, J. P., Hendricks, K. V., Eckardt, R. E., Stanton, C . L., and Page, R. C., Arch. I d . Hug. Occup. Med., 4, 3 2 5 (1951). (8) Hueper, W. C., “Occupational Tumors and Allied Diseases,” Chap. 11, Springfield, Ill., Charles C Thomas, 1942. ( 9 ) Ibid., pp. 1 4 6 5 3 . (10) Setlllll, K., and Ekwall, P., Nature, 166, 188 (1950). (11) Smith, W. E., Sunderland, D. A,, and Sugiura, K., Brdi. End,. Hug. Occup. Med., 4, 2 9 9 (1951). (12) Twort, C. C., and Twort, J. M., J. Ind. Hyg., 13, 204 (1931). (13) Wanless, G. G., Eby, L. T., and Rehner, John, Jr., Anal. Chem.,. 23, 563 (1951). RECEIVED for review Deoember 19. 1951.
ACCEPTEDApril 2 6 . 1952..
Granular Adsorbents for Sugar -
Refining SOME FACTORS AFFECTING POROSITY AND ACTIVITY IN SERVICE ELLIOTT P. BARRETT R a u g h and Sons Co., Philadelphia,Pa., and M e l l o n I n s t i t u t e , P i t t s b u r g h , P a .
L. G. JOYNER AND P. P. HALENDA Mellon I n s t i t u t e , P i t t s b u r g h , P a .
T
HE gradual decrease in color- and ash-removal power of bone char with increasing length of service in sugar refining is a familiar phenomenon, but the mechanisms t h a t operate t o produce the deterioration in activity have not been adequately studied. I n conjunction with a description of the development and refinery scale testing of a synthetic granular adsorbent for sugar refining, called Synthad (2-38, Barrett, Brown, and Oleck ( a ) studied the changes in distribution of pore volume and area of the synthetic and natural chars in service. T h e activity, indicated by either color- or ash-removal power, decreased less rapidly in proportion t o initial activity than area decreased in proportion t o initial area. In other words, activity per unit area increased in service. An exception t o this was displayed by Synthad in the early cycles, as it gained area instead of beginning to lose it immediately. On the baais of these results and those of t h e pore volume and area distribution studies, i t was concluded t h a t there was no evidence t o indicate t h a t any of the pores in either adsorbent were too small t o function in the removal of impurities from sugar liquors. The loss in small pore ( - 100 A. radius) area was shown
to occur by two mechanisms--filling of the pores with adsorbed impurities and the growth of hydroxyapatite crystallites. The latter phenomenon, apparently, would effectively reduce the area by a redistribution of the interstices between crystallites, these interstices constituting the so-ealled pores. Reduction in area b y this mechanism will occur whenever environmental conditions favor crystallite growth regardless of whether or not the char has been contacted with impure sugar liquors, Le., whether the char has done any “wo&” or not. This paper describes the results of laboratory scale experiments calculated t o assist in understanding some of the results reported in a previous paper (2)and t o determine the effect of various conditions on the rate of crystallite growth, and, consequently, on t h e rate of loss of area. The d a t a are presented in t h e form of pore volume and area distribution curves constructed from nitrogen desorption isotherms by the method of Barrett and Joyner (8). Prior t o determining the desorption isotherms, t h e samples were outgassed for 1 hour at 200 C. under a vacuum of approximately 2 microns. To test the reliability of the results calculated from the nitrogen desorption isotherms, an independent approach
