2220
INDUSTRIAL AND ENGINEERING CHEMISTRY
N-( 2,2-Difluoroethyl)-N-( 2-hydroxgpropy1)aniline. A mixture of 23.5 grams (0.15 mole) of N-(2,2-difluoroethyl)anilineand 10.3 grams (0.15 mole) of propylene oxide dissolved in 10 ml. of ethanol was sealed in a Carius tube. The mixture Tas then agitated a t 200' C. for 18 hours in a rocking autoclave. The mixture was distilled to obtain 18 grams (63.0%) of the desired product, b.p. 135-137" C. (6 mm.), ng 1.5187. Calculated for C11HljF21iO: C, 61.40; H, 6.98; N, 6.51. Found: C, 61.05; H, 6.99; N, 6.54. N-(3,3-Difluorobutyl)-N-(2,3-dihydroxypropyl)aniline. A mixture of 9 grams (0.049 mole) of fiT-(3,3-difluorobuty1)aniline and 4.6 g r a m (0.055 mole) of sodium bicarbonate was placed in a three-necked flask fitted with a mechanical stirrer. The mixture was stirred and heated a t 145' C. while 6.1 grams (0.055 mole) of l-chloro-2,3-propanediolwas added dropwise during 30 min. and carbon dioxide was evolved. Heating was continued for 6 hours more. The reaction mixture became very viscous and dark. The product Kas treated with 50 ml. of water and 50 ml. of benzene, then stirred. The benzene layer was separated, and the aqueous layer was extracted with three 25-ml. portions of benzene. The benzene extracts were combined and steam distilled to obtain 12.1 grams (95.3%) of a viscous, brown liquid. This crude product was distilled in a molecular still a t 22 p with the bath temperature at 110" to 125" C. t o obtain an oil, n g 1.5230. Calculated for C I ~ H I ~ F & O C, ~ : 60.2; H, 7.4; X, 5.4. Found C, 60.1; H, 7.3; N, 5.3. N-(2,2-Difluoroethyl)-N-(3-hydroxgpropyl)aniline. A mixture of 47 grams (0.3 mole) of Ar-(2,2-difluoroethyl)anilineand 30.5 grams (0.36 mole) of sodium bicarbonate was stirred and heated a t 135" C. uhile adding 46.4 grams (0.36 mole) of 3bromo-I-propanol during one half hour. After the addition, stirring and heating at 140' C. were continued for 6 hours. The product distilled at 117' to 121' C. (1.5 mm.), n g 1.5234. The yield was 9.5 grams (15.0%). Calculated for CllHlbFQNO: C, 61.40; H, 6.98: N, 6.51. Found: C, 61.05, H, 6.81; Ii,6.50. SUMMARY
,4new class of compounds, N-fluoroalkylanilines, was prepared by treating anilines with various fluoroalkyl halides. Some of these fluoroalkyl halides are new compounds; others are known compounds, and some were prepared by improved methods. The fluoroalkyl groups introduced into the anilines by alkylation were 2-fluoroethyl; 2,2-difluoroethyl; 2,2,2-trifluoroethyl; 2,2-
difluoropropyl; 3,3-difluoropropyl; 3,3,3-trifluoropropyl; 3 , s difluorobut,yl; and 4,4-difluoropentyl. -Y-( lH,Iff-heptafluorobut,yl)aniline, however, was prepared by lithium aluminum hydride reduction of heptafluorobutyranilide. These S-fluoronlkylanilines were hydroxyalkylated Eith et'hylene oxide, propylene oxide, 3-hromo-l-propanol, and 3chloro-1,2-propanediol to obtain the corresponding N-fluoroalkyl-iv-hvdrosyalkyl~nilines. The properties of the S-fluoroalkylanilines and of tho N fluoroalkyl- hydroxyalk alkyl anilines are greatly influenced by the introduction of fluorine, The effect is dependent on the degree of fluorine substitution and on the position of this substitution in the alkyl group. The effect appears in both the chemical and the physical properties. LITERATURE CITED
Benning, A. F. (to E. I. d u P o n t de Nemours & Co.), U. 3. P a t e n t 2,230,925 (1941). Dickey. J. €3. (to Eastman Kodak Co.), E. S. Patents (a) 2,516,106; (b) 2,516.107; ( e ) 2,516,302; (d) 2,516,303 (1950); ( e ) 2,594,297: (f) 2.815.013; (g) 2,615,014; (h) 2,618,630 (1952). Dickey. J. B., and associates, 1x11. Erc,. CHIN., 45, 1730 (1953). Haszeldine. R.N., J . Chem. Soc., 1950, p , 2789. TTenne, A. L.. and Flanagan, J. V., J . A m . C'hem. Soc.. 65, 2362 (1943). Henne, A. L., and Haeckl, F. W., Ihid., 63, 2692 (1941). Henne, A . L., and Hinkamp, J. B.. Ibid. 67, 1194 (1945). Henne, A. L., and 3Iidgley. T., Jr.. Ihid.,58, 886 (1936). Heniie, A. L., and Renoll, X a r y K., Ibid..58, 890 (1936). Ibid.,59,2431 (1937). Henne, A. L., and Ruh, R. P., Ibid., 70, 1026 (1948). L., and Waalkes, T . P.. Ibid.,68, 496 (1946). Henne, A. L., and Whaley, A. l I . , I b i d . , 64, 1157 (1932). Henne, -1. L., and Zimmerschied, W. J.,Ibid.. 69, 282 (1947). McBee, E. T., and associates, INI).ENG.CHEM.,39,409 (1947). l I c B e e , E. T., and associates. J . Ana. Chem. Soc., 69, 944 (1947). McBee, E. T..and Hausch, W. R., IND. ENG.CHEX, 39, 418 (1947). Muskat, I. E., and Nort,hrup, H. E.. J . Am. C h ~ mSoc., . 52,4061 (1930). Renoll, Mary W . , Ibid.,64, 1115 (1942). Swarts, E"., (a) Bidl. classe ,sei.. Acad. rov. Relg., GO (1909); Rec. trao. citim., 28, 166 (1909); (b) Bull. cZasse sei., Read. roy. Uelg., 576 (1911). Towne. E. B., and Dirkey, J. €3. (to Eastman Kodak Co.), U. 8. P a t e n t 2,500,218 (1950). RECEIVED for reviex February 13, 1954. ACCEPTEDJune 14, 1954. Presented as part of the Syinposiuin on Fluwine before the Diviuiun of Industrial and Engineering Chemktry a t the 124th Meeting of the AXERIOAN CHEXICAL Q O C I E T T , Chioago, Ill.. September 19%.
Catalvsis bv Co J
d
Vol. 46. No. 10
exes
DRYING OF LINSEED OIL A. C. ZETTLEMOYER AND RAYMQND R. MYERS National Printing Ink Research Institute, Lehigh University, Bethlehem, Pa.
E
VEIt since the first accidental inclusion of oxides of lead in
linseed oil, metal salts have been employed as drying catalysts in the paint, varnish, and ink industry. In the form of soluble salts of organic acids, the divalent ions of manganese, cobalt, and lead have dominated drier technology for decades. The supremacy of these conventional catalysts was challenged first by the discovery that certain complex compounds of iron were of biological importance because of their ability selectively t o catalyee oxidations under mild conditions-the role of hemo-
globin in animal mctabolism ( 7 ) is an example. In view of the fact that iron normally is an oxidation catalyst only a t high temperature, the conclusion that the type of bond peculiar to these complex molecules was responsible for the greater activitv led early experimenters to a study of relat,ed compounds a8 driers for inks. Becauve of the similarity between biochemical oxidations and the drying of linseed oil, the use of hemin in the oxidation of drying oils was tried its early as thirty years ago by Robinson (9) in a
October 1954
INDUSTRIAL AND ENGINEERING CHEMISTRY
series of successful experiments in which the catalyst was present in the form of an aqueous emulsion. An extension of the work by Haurowitz and coworkers ( 5 ) ,in which homogeneous solutions were used, was unsuccessful a t first. Later, these same workers ( 4 ) found that hemin catalyzed the oxidation of linoleic and linolenic acids but was destroyed in the process. Zettlemoyer and Nace ( I f ) produced an oil-soluble hemin oleate which worked well as a catalyst and n a s particularly active in pigmented systems. Unfortunately, the active components of blood constitute less than 2% of the solids content which, combined with recovery difficulties has placed a prohibitive price on hemin as a catalyst. These considerations have led to a search for a simpler solution, in which complexes were synthesized around the central metal. In 1941 Panimon and associates (8) found a synthetic iron-ophenanthroline complex which would influence biochemical processes in a fashion similar to the effect of hemin. This discovery suggested the possibility of effecting economies in driers through the discovery of a less expensive amine, honwer, the first 10-year endeavor resulted only in the discovery by Wheeler (10) that o-phenanthroline and a,a’-dipyridyl could be employed as drier accelerators in pigmented systems merely by being added to a system already catalyzed by conventional driers. Prior to Wheeler’s discovery, Nicholson ( 7 ) had learned of the usefulness of certain coordinated metal soaps but had attributed their activity to reduced adsorption of the’metal by the pigment. The cost of these early organic adducts was high, however. While certain restrictions must be placed on the use of complexes, they can he regarded as the first step in a new drier technology because of their accelerating and stabilizing qualities. Chief among these limitations are the facts that the most important conventional drier (cobalt) has resisted improvement via complexing in unpigmented systems and that the relationship between a complex drier and the components of an oleoresinous system is less predictable than when conventional driers are used. Occasionally a product of different physical characteristics has resulted from the use of a complex. The study of complexes in this laboratory has produced information of practical value-certain amines were found that prevented excessive loss of drying on aging in certain pigmented vehicles, and the efficiency of iron in the form of its naturally occurring complexes had been reproduced in synthetic materials of comparatively simple structure. The reasons for both phenomena remained unexplained. A fundamental approach to the problem of explaining activation via the complexing route was chosen for further study. The work was done in two distinct phases which consisted of: 1. Addition of amines and other drier aids to metal-catalyzed linseed oil 2. Preparation of amine complexes of mctals and thrir evaluation as driers 1
This paper covers only the portion of the work done with amine additives. Results of the complexing studies will be reported later. Sufficient work has been done in this laboratory and elsewhere ( 7 ) to prove that the two techniques produce similar results. EQUIPMENT
Wholesale evaluation of films containing various types of driers required an apparatus capable of recording drying times while unattended. The Gardner drying time recorder ( 3 ) was employed throughout this work. It consists of a Bakelite dolly 29 x 4 inches equipped with casters a t either end. and directed by a metal guide to progress horizontally a t a rate of three centimeters per hour by means of a rack engagement to a synchronous motor. Clamped to each dolly is a pair of plate glass slides, 27 x 1 inch, containing the film to be evaluated, and passing b e neath a 5-inch gear whose impression is left on the film.
2221
Constant temperature, light intensity, and air flow were maintained by a plywood cabinet equipped with a 15-watt fluorescent lamp mounted horizontally and transversely over the dolly at a point preceding its passage under the gear, a de Khotinsky thermoregulator assembled in series with a 750-watt radiant heater mounted inside a perforated metal container. Air was supplied at a linear velocity of 160 centimeters per minute by a small centrifugal blower mounted outside the cabinet. While humidity was not constant, control f i l m of uncomplexed metal were used for daily cornparkon. Films were spread evenly and reproducibly by means of a Gardner Film+-Graph which consists of a stainless steel cylinder in one end of which is milled a groove 0.001-inch deep. Drying times were read in terms of the distance traveled by the glass slide. MATERIALS AND STANDARDS
Cobalt, manganese, lead, and iron naphthenate driers were employed throughout this work; nickel catalyst was prepared in the form of the %ethyl hexoate. Of the 160 amines tested, the chief sources were Carbide %I Carbon Chemicals Division, Rohm & Haas, American Cyanamid, Tennessee Eastman, Monsanto Chemical Co., Armour Co., Sharples Chemical Co., Alrose Co., and Harshaw Chemical Co. The important class of Schiff’s bases was prepared in the laboratory according to directions given in “Inorganic Synthesis” (1). The linseed oil was purchased in a 50-gallon lot from ArcherDaniels-Midland Co., and met ASTM Specification D 234-48. A standard concentration given in Table I was set for each metal studied.
TABLE I.
STANDARD
METALCOSCENTRATIONS Std. Conon., Wt. yo Metal 0.03 0.05
Metal Cobalt Manganese Iron Nickel Lead
0.05 0.05
0.30
PROCEDURE
The test method consisted of reading the times of the characteristic gear impressions made upon the film as it dried. The criterion of drying, the so-called drop time, occurred when the thread of material breaking away from the departing tooth of the wheel had become sufficiently rigid to prevent a small drop from forming in the center of the impression. This criterion has been referred to as set-to-touch time in Gardner and Sward ( 3 ) . Whenever possible, solutions containing 0.04 moles of amine per liter of ASTM linseed oil were made. Stock solutions of each metal in the same vehicle were prepared a t twice the standard concentration, resulting in a solution 0.006 molar in the case of cobalt, and proportionately higher in the case of the other metals. Blending was done in 12-ml. tubes by pipeting exactly 5 ml. of the metal stock solution, a quantity of amine solution calculated to produce a 2: 1 molar ratio of amine to metal, and the amount of linseed oil required to bring the total volume to 10 ml. Cobalt analyses were performed according to the method developed by Gamble (,%’)in these laboratories, in which the metal was removed from the organic medium by successive extractions with concentrated hydrochloric acid, and the blue coloration of cobalt chloride (CoCls) measured with a Coleman spectrophotometer. Manganese analyses were performed spectrophotometrically by a similar method and also by conversion to the sulfate and ignition. RESULTS
With the adoption of 0.03% cobalt as the standard, results were obtained in terms 86) nearly free of uncontrolled variables
2222
INDUSTRIAL AND
E N G I N E E R I N G CHEMISTRY
possible by dividing the drying time of 0.03% uncomplesed cobalt by the time required for the complex formulation to dry under identical conditions, a figure usually in the vicinity of unity was obtained which mas expressible t o the second decimal point. I n the case of manganese, evaluations based on a cobalt control (0.03y0conventional cobalt drier) werr used frequently in addition t o those based on 0.05% manganese. Results were expressed in ternis of percentage improvement of the metal in question; in the case of manganese, a 100ycimprovement represented the activity of the cobalt standard, while with iron a 3007, improvement was necessary t,o gain parity with cobalt. By operation a t a range of from 2 to 6 moles of amine per metal atom, the observation was made that effects were as measurable at the lower concentration as st the high ones; consequentlv, a 2 : 1 mdar ratio of amine to metal was employed. This ratio coincided with a characteristic coordination number of the metals employed. Complete results covering the 160 amines tested are too lengthy t o be reported here; the most pertinent findings are summarized:
Vol. 46, No. 10
a8
1: Catalysis by cobalt was unimproved beyond 20% by any a m n e . Those which showed some activity were benzyltriinethylammonium butoxide, diethylethanolamine, diphenylguanidine, 2-methyl-6-aminopyridine, and sym-diphenylcarbazide. This slight improvement is believed to result from neutralization of antioxidants in the vehicle; a t any rate, it was n o t considered significant. 2 . Cobalt catalysis was retarded extensively by a t least 18 of the amines. The most potent retarders were o-phenanthroline, substituted diphenylamines, tetrabutylethylenediaminetetrapcetate, N,1V’-disalicylalethylenediamine, salicylaldoxinie, ammoguanidine, several quaternaries, and diamines in general. o-Phenanthroline had been reported as a cobalt accelerator by Nicholson (7), but this finding was not confirmed in unpigmented systems. 3. Manganese catalysis v a s accelerated to 1S070 by 44 amines, sometimes to a point a t which it was more effective than cobalt. Of t.he top 24, a t least 14 were difunctional heterocyclic amines, and two others mere characterized by the presence of a doubly bound nitrogen conjugated with an aromatic ring; those which did not fit into this pattern were highly substituted with large groups. Table I1 lists the 21 most active amines roughly in the order of decreasing potency.
RATORS
Additive 1. K,.V’-Disalicylalethylenediaminit 2. ~‘,N’-Disalicylal-u-phenyleIie[~ia~iiiii~ 3. o-Phenanthroline 4. Dicyclohexylamine a. IIexamethy1enetetrai:iine 6. Cyclohexylamine 7. N,N’-Dibenraletli~lenrdianiine 8. Picolinio acid
9.
10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21.
o-Dimethylaininoinetli~l-p-octyiphenol
(DMP 18) o-DimethylsminometIiyl-p-b~it~l~,heIiol (DMP 14) Dicyandiamide a-Naphthylamine Methyldiethanolamine 2.~~ethyl-6-aminopyridine Lauryl pyridinium chloride 2-(2-Dimethylaminoethylalnino)pyridine Tris(hydroxymet1iyl)aminoethane A’-Dicyanoethylbenaenesulfona~uide
Quinaldine 2-Vinylpyridine ?-Aminopyrimidine 22. Tdphenylguanidine 23. -4c~toacetanilide
24.
dl-iilanine
Actirity, 111ig:ove0.03 Co inent, = 1.0 “0 1.40
1.35 1.35
1.10 1.00
1.00
180 170
170 120
100
0.95 0.93
100 100 i)0 90
0.95
90
0.90 0.90 0.90
80 80 80 60
1.00
0.80 0.80
0.75 0.73 0.70 0.70
GO
50
50
40 40
0.70
40 40 10 40
0.70
’10
0.70 0.70 0.70
4. Manganese was retarded by strongly h s i r diamines, suhstituted phenylenediamines, diphenylamine, dietliS.leiietrianiine, 2-cystine, tetrabutylethylenediaininetetrnacetate> and sulfurbearing amines. 5. Iron was accelerated to 300% by many of the amines found $ i y e z i t h manganese. Especially potent were those listed in l a w e 111, , 6. Kone of the amines accelerated nickel catalysis to the poirit at which that nietal could be coilsidered a primnry drier.
TABLE III. IROU .ACCELERATORS rlotivity, 0.03 Co 1.
2. 3.
p.
a. 6. 7. 5. !I. 10.
Additive o-Phenanthroline iV,ATf-Disalic3,1alethyienedaiiiine Triphenylguanidine Pyridine Allylamine o-Phenylenediamine p-Bromoacetanilide DMP 14 Tribenzylarnine Quinaldine
Additive 1. ~,~‘-Disalicyl-o-phenyleni.dia~iiine 2 . a,a’-Dipyridyl 3. 2-Aminoethyl sulfuric acid 4. sum-Diphenylcarbazide 5. Sitro-o-pl-ienant!iroline 6 . Salicvlaldoiime 7. Urea8. p-Bromoacetanilide 9 . dl-Leucine 10. \ietiiylacetylurea
Im~xova-
ment,
= 1.0
Dic
0.60
300
0.46
0.40 0.35 0.35 0.36
0.35 0.35 0.35 0.45
210 170
130
130 130
130 130 130
130
Activity. 0.03 Co
Improvement,
0.70
50 40 30
= 1.0
0.65 0.60 0.53
0.55
0.55
0.50 0.60 0 50 0 50
$6
20 20 “0 10 10 10 10
Some assistance T Y W afforded by o-phenanthroline, a,a’-dipyridyl, DM P 14, K,A”-disalicylalethylenediamine,certain quaternaiirs, triphenylguanidine, and 2-motd~yl-6-aminopyridine. 7 . Lead catalysis was accelerated to a maximum of 50% bh the amines listed in Table IT7. INTERPRETATION OF RESULTS
I n most cases additiyes effective with manganese, iron, and nickel driers possessed a resonant structure with taro opposing effects :
1. The basicity of the amine was reduced by the resonance. 2. The coordinatiIig t,endency of the amine was restored partially by structural effec:t,sin the molecule to compens:itc for the reduced basicity. Generally, the coordinating tendency of an amine-i.e., its willingness to supply the two electrons necessary for the formation of the type of bond found in complexes-can be correlated with its basicity. There is a special case, however, in which coordinating tendency is much greater than can be attributed to hasicity, alone-that case is when two or more functional groups containing nitrogen or oxygen exist in the same molecule. Such bifunctional compounds appear t g form complexes with metals more readily than anticipated from their basicity, because the entrance of one group into combination with the metal induces the second group to form a similar bond. The resulting chelate is harder t o break down than arc two individually held monoamines. The desired structure is found in the central groupings which are Characterized not only by the formation of 5- or 6,rnembered chelate rings with the metal but also by resonance. The doubly bound, resonating nitrogen appears to be the keystone in the construction of a chelate which is less stable than one involving %: nonresonating molecule. I n this category are found o-phenanthroline, picolinic acid, the Schiff bases, and various amino ac In the case of o-phenanthroline the bifunctional character is supplied by the two nitrogens; with the amino acids the carboxyl group plays the role of the second ‘center of coordination. The Schiff bases are interest’ing in that they provide an economical means of upgrading manganese and iron. In contrast with the general efficacy of reasonant difunctional structures in improving manganese and iron is the retarding effect of aliphatic diamines. These amines are strongly basic and would be expected to fill two coordination positions of the metal ion, in which case four extra electrons have been introduced into the cluster of electrons around the metal. By comparison, the resonant diamines operate through only one nitrogen
INDUSTRIAL AND ENGINEERING CHEMISTRY
October 1954
a t a given time. That is, when one of the functional groups is acting as an electron donor the rest of the molecule is so highly polarized that the second functional group cannot act in the same capacity a t the same time. Reversal of the polarity during the resonance cycle causes the two groups to change roles. The net result of coordination involving B resohating diamine is an electronic configuration resembling a monoammine in a structure possessing the steric features of the more favored diammine. The possibilities are compared with the electronic configuration of uncomplexed cobalt in Table V, in which each column represents an orbit, or more appropriately, an orbital, capable of holding two electrons.
In some cases manganese catalysis was improved to the point a t which it was superior to cobalt catalysis; in general, the greatest improvement occurred with diamines or other difunctional amines possessing resonant structures and by certain amines of large steric requirements. Iron was susceptible to improvement by the ssme types of amines and to be a greater extent in proportion to its original activity than was manganese. ACKNOWLEDGMENT
Laboratory assistance in testing the amines was provided by Carolyn Moore and Gilbert Epstein. The work was supported by the Harshaw Chemical Co. and Advance Solvents 8: Chemical Co., and was conducted as a cooperative proiect a t the National Printing Ink Research Institute a t Lehigh University.
COSFIGURATIONS OF DIAYMINES TABLE Ir. ELECT~OSIC
co++
M n (A)+++* CdA) Mn(A);++’ Co(A)e
.. xx
xx xx xx
.. .. ..
xx
xx
3d
.. .. ..
LITERATURE CITED
48
.* ..
..
..
From Table V the observation can be made that a manganese monoammine Mn(A) *+ resembles uncomplexed cobalt in ita electronic configuration. That is, there are 7 electrons in the 5 orbitals. The apparent role of the resonance is to permit the monoammine rather than the diammine configuration Mn( A)z + * to be attained. In the case of cobalt, any complex species contains more electrons than does cobalt, itselfAn extension of these views appears in the accompanying paper (6) in which the definite pattern described by the various accelerators and retarders is used as the basis for an electronic explanation of the catalysis of linseed oil drying. SUMMARY
Economical replacement of conventional drying catalysts for oleoresinous vehicles by metal complexes has been demonstrated. A means has been shown whereby cobalt can be replaced by manganese and iron in conjunction with a suitable amine additive.
2223
(1) Audrieth, L. F., editor, “Inorganic Syntheses,” Val. 111, p. 196, McGraw-Hill, New York, N. Y.,1950. (2) Gamble, E.,Natl. Printing Inlc Research Inst. Bd.,16, Sept. 1, 1949. (3) Gardner, H. A, and Sward, G. G., “Physical and Chemical Examination of Paints, Varnishes, Lacquers and Colors,”
11th Ed., p. 157,Inst. of Paint and Varnish Research, Henry A. Gardner Laboratory, Bethesda, Md., 1950. (4) Haurowitz, F.,and Schwerin, P., Enzymologia, 9, 193 (1941). (5) Haurowitz, F.,Schwerin, P., and Yenson, M. M., J . Bid. Chem., 140, 353 (1941). (6) Myers, R. R., and Zettlemoyer, A. C., IND.ENG.CHEM.,46, 222.7 llS54i ~. ~, . .
(7) Nicholson, D.G., Ibid., 34, 1175 (1942). (8) Panimon, F., Horwitt, hl. K., and Gerard, R. W., J. CeUlllar Comp. Physiol., 17, 1, 17 (1941). (9) Robinson, M.E.,B i o c h a . J., 18, 255 (1924). (10) Wheeler, G. K., U. 9. Patents 2,926,758 (Oct. 24, 1950) and 2,565,897 (Aug. 28, 1951); Im. ENG. CHEM.,39, 1115 (1947). (11) Zettlemoyer, A. C.,and Nace, D. M., IND.ENG.CHEM.,42, 491 (1950). RWEIVED for review November 3, 1953. ACCEPTED June 8, 1954. Presented before the Division of Paint, Varnish, and Plastios Chemistry, a t the 124th Meeting, ACS, September 1953. From a dissertation submitted by Raymond R. Myers to the graduate school of Lehigh University in partiel fulfillment of the requirements for the degree of doctor of philosophy, June 1952
(Catalysis by Complexes)
ELECTRONIC EXPLANATION OF CATALYSIS OF LINSEED OIL DRYING RAYMOXD R. MYERS AND A. C. ZETTLEMOYER
s
4 RAL phenomena were observed to occur repeatedly in the amme evaluation program reported in the first paper of this series ( 5 ) . On the basis of observations on over 160 amines, a hypothesis has been advanced to explain the role of the electronic configurations of the metals and metal ammines used as catalysts for the process. By means of the hypothesis a path has been cleared toward the development of an electronic theory of catalysis involving covalently and ionically bound metals and metal complexes. The explanation of the catalysis of linseed oil drying has been used as the basis for the investigation of complexes as driers, and conceivably can be used in directing the course of action in numerous complexing studies. The metal-catalyzed drying of linseed oil has been selected for study owing to the wealth of data available. Many investigators have been concerned with the mechanism of oxidation and have contributed to its elucidation. Ample proof has been offered by Farmer (2)that the first step in drying
is oxidation of the vehicle. This oxidation step appears t o be the rate-determining step, inasmuch as an active oxygen is needed to abstract a hydrogen atom from the alpha carbon. The function of the drier appears to be an activation of the oxygen so that it will go about this task more efficiently. Regardless of the mechanism of autoxidation of the vehicle, the chief concern of this study centered around the reaction of the metal with oxygen; this reaction must be neither so energetic that the metal-oxygen bond will not be broken easily, nor so weak that the catalyst is not effective. The effectiveness of a catalyst can be measured by physical changes in the vehicle, aa manifested by the Gardner drop time ( 3 ) which is measured by the passage of a gear across the film during drying. Essentially this criterion is an indication that the viscosity of the vehicle has increased to a point a t which i t indicates a structure sufficiently strong to prevent the formation of threads between the moving gear and the film.
2224
INDUSTRIAL AND ENGINEERING CHEMISTRY GENERAL OBSERVATIONS
Facts which assisted in the development of the hypothesis, or which must be explained by it, consiat of:
1. The electronic configurations of the various metal ions, established from spectroscopic data, are as depicted in Figure 1. Work currently in progress indicates that the metals are not in the ionic state, but are covalently bound. This situation does not interfere with the present discussion which has been based on the more conventional ionic representation. 2. I n the periodic arrangement of the elements, a general tendency toward orbital symmetry is noted with the result that in the case of the d shell there are three favored configurations; no d electrons, 5 electrons in 5 different orbitals, or 10 electrons in the 5 orbitals ( 1 ) . The 4s electrons present in the native metals listed in Figure 1 do not enter into the discussion a t this time; they may or may not be present, depending on the nature of the salt. 3. Peculiar redox characteristics resulting from the orbital symmetry tendency of the 3d electrons are that ferric ion is more stable than ferrous, that copper is the only element of this group Tvhich exists in the monovalent state, that the manganic ( f 3 ) ion is converted readily to a $2 or +7 valence, and that the +4 valence of nickel is preferred over the +3 valence. 4. Complexing changes the redox characteristics of the ions in the group considered here. For example, the redox potential of the cobaltous-cobaltic couple is reduced from 1.8 to -0.8 volts by cyanide; the ferrous-ferric couple is increased from 0.74 t o 1.14 volts by o-phenanthroline. 5. Strongly basic diamines reduce the efficacy of cobalt and manganese catalysts. Sterically hindered amines and compounds containing nitrogen in a resonating system are more likely to enhance the efficacy of manganese than do aliphatic amines. 6. The enhancement of the activity of cobalt via the complexing route occurs only in certain pigmented systems (6). ELECTRONIC MECHAMSM
At present the discussion is concerned with uncomplexed ions. Figure 1 is the conventional representation of the arrangement of the outermost electrons of a few of the transition elements in the third period. The large circles represent orbitals, the energy equivalent of the more readily visualized concept of a t o m i c orbits; each 30 4s orbital can contain from ~ ~ + 00000 2 0 zero to two electrons, and FE+3 00000 0 may share one electron with F E + ~ 00000 0 the orbit,al of another atom. Bond f o r m a t i o n o c c u r s Cof2 OOooo either when a sharing of NI c2 00oo0 0 electrons takes place or C"" 00000 0 when an electron is lost Figure 1. Electronic from the orbit,al in which Configurations it is found in the neutral a t o m . R e g a r d l e s s of which type of bond is formed, if the participant lies to the right of the metal in t,he periodic table the metal is considered to be oxidized in the process. This view serves as the basis for this discussion. From Figure 1 t,he ion which should make the most active oxidation catalyst will be predicted from a consideration of the probability of establishing a covalent homopolar linkage with oxygen. The statistical advantage of certain metals will then be shown to be counteracted by energetic features involving the 3d electrons. Statistically, the manganous ion has a 25% better chance of attracting an oxygen molecule than does it's nearest competitor, ferrous ion. This advantage results from the five half-filled d orbitals of manganese compared with the four such orbitals in ferrous iron. Ferric iron has the same statistical advantage as manganese but would have t o be oxidized above a valence of three to permit it to be a catalyst since an electron is shared with oxygen. Ferrous iron enjoys a statistical advantage over the remaining metal ions in the transition group, and might be ex-
Vol. 46, No. 10
pected to be a strong catalyst as a result. Consequently, one might predict that bobh manganese and iron would produce better drying catalysts than the remaining metals. However, experience teaches that neither manganese nor iron is as effective as cobalt which possesses only three half-filled 3d orbitals. Clearly, the explanat'ion involves some effect not yet considered. Reference to the uniform atomic plan conceived by nature supplies the answer. I n the assemblage of elements that constitutes our periodic table one can find only two cases in which an s electron is not endowed with less energy than a d electron of the next lower principal quantum number. These cases are when the d shells are one electron short of being half-filled as in the case of chromium and completely filled as in the case of copper. The tendency t9ward orbital symmetry gives rise to these favored configurations, the att,ainment of which results in lower energy than one would expect for the metal in quest,ion. These energetic features are superimposed on the statistical factors governing the number of reversible unions with oxygen. As a corollary, one might expect to find the energetic requirements of the addit'ion of an elect'ron to a favored configuration somewhat higher than the normal energy associated with a 3d orbital. The reason for the reluctance of mnganese to assume a + 3 valence, despite its st,atistical advantage, is bhat the favored configuration of 5 electrons in 5 orbitals must be deptroyed in the process. This is true regardless of whether oxidat,ion is viewed as the loss of an electron by the metal to the oxygen or as the formation of a new homopolar bond. The passivity of iron as a drier. and its extreme willingness to enter the trivalent state are manifestations of the same phenomenon. Iron has both the statistical and energetic features which are conducive t o union with oxygen, but the requirement that t,he process be reversible has not been met. in this case; ferric iron, once formed, is difficult to reduce, because it has attained the favored configurat,ion of 5 electrons in 5 orbitals. Cobalt offers the right, combination of statistical and energetic features. Not only does the cobaltous ion possess three degenerate forms of the one-electron orbital, but it is characterized further by the fact that a t no time during the process is a favored configuration approached. Consequently, this "inertial energy" is not required in the acceptance of oxygen nor in its subsequent loss to the vehicle. On reaching the configuration of nickel the two effects work together to render Ni++ impotent as an oxidizing catalyst, as contrasted with niclrel-catalyzed oxidations and reductions in which the 4s electrons participate. Not only has the statistical factor been reduced to two, but also the filling in of the 3d orbitals has progressed to the point a t which the attachment, of one oxygen with its electron results in the configuration, 3d8. Since this configuration is sufficient in the case of copper t o induce the capture of the tenth electron, the surmise t'hat nickel, with a nuclear charge only less than copper, can take on a second oxygen with a minimum expenditure of energy ie logical. The obscrvation that nickel is unique in its avoidance of a +3 valence results from the 3dQConfiguration which it Tvould be forced to assume. EFFECT OF COMPLEXIZG
If a may can be found in TT-hich the electronic energy requirements of the 3cl orbitals can be lowered simultaneously with the ret,ainment of a favorable statistical advantage, drastic increases in the number of reversible unions with oxygen can occur. This effect is exactly xvhat happens in complexing, particularly when a deficiency of amine is employed so that only the partially coordinated species are prevalent ( 5 ) . I n the case of cobalt, the insertion of one iiit'rogen atom with its pair of electrons destroys the active cobalt configuration and tends to reduce the efficiency of the catalyst. If two nitrogene are introduced into Co++, the result will be a configuration in which one lone electron will
October 1954
INDUSTRIAL AND ENGINEERING CHEMISTRY
occupy either a 4s orbital in the event the cobalt is ionically bond or a 471 orbital in covalently bound salts. I n either case the Co++ will be susceptible to irreversible ionization. This lability of the single electron accounts for the stability of trivalent cobalt when complexed. Further complexing ultimately forces this electron into the 5s orbital where it has even more tendency t o ionize. Having seen that the motion of an electron in and out of an orbital is facilitated if two filled orbitals are present as a bulwark against the attainment of a favored configuration, one can predict that the addition of two electrons to manganese via complexing will remove the electronic energy barrier operating against oxidation of the uncomplexed metal, moreover, the energy requirements involved in the gain and loss of electrons will be less than in the case of cobalt owing to the lower nuclear charge on the manganese ion. The effect of the nuclear charge cannot be emphasized too highly. Not only does i t influence the activation energy required for the formation of the activated state, but it is the driving force behind the attainment of a favored configuration. I n the case of manganese versus cobalt this means that not only will the monoammine possess greater oxidizing potential than cobalt, but also the diammine, with its 3dDconfiguration, will not possess nearly so much reserve affinity for the tenth electron as does the isoelectronic copper ion which is four units heavier. Consequently, the diammine may have some oxidizing ability although it may be inferior to the monoammine. The effect of the nuclear charge is to permit the manganese monoammine to be a much better catalyst than uncomplexed cobalt. It is the cobalt configuration, rnther than cobalt, itself, which is believed active in catalysis. Up to a point, the characteristics sought in a complex catalyst are the attainment of the cobalt configuration and a lowering of the energy requirements for movement of electrons in and out of the orbitals. An interesting corollary of this hypothesis is that a vanadium diammine should prove to be an even more potent catalyst than the postulated manganese monoammine. Scandium, too, is of interest as a candidate for future complexing studies because the triammine of that metal is isoelectronic with cobalt. These possibilities currently are being investigated. Lowering of the energy requirements of the orbitals cannot be carried on indefinitely, for complexing cannot be carried past the next rare gas structure. Even if the limiting case, in which no energy is required, were attainable the catalyst probably would be inoperative, for the absence of a catalyst can be viewed as an example of the limiting case.
which were found active with manganese in the first paper of this stress. One may portray the metal catalyst in the role of free radical fxmer. The knowledge that molecular oxygen prefers the electronic structure
..
..
0:o. .. ..
to the more conventional11 written form
.. ..
0 . .: :o .. can be added to the concept of covalent bond formation with cobalt through one of the 3d orbitals, in which case the activated complex is written Co++:O:O. .. .. thus showing unequivocally the free radical nature of the spearhead oxygenbatom. After the hydrogen atom, or the linseed oil free radical, depending upon one’s concept of oxidation, is incorporated into this structure, the cobalt-oxygen bond is broken and the catalyst released for further oxidation. An explanation of the ability of iron to function efficiently as a drying catalyst a t high temperatures may be in the thermodynamics of the decomposition of the structure
..
F e + + :0: . . 0: .. H to release the ferrous ion. Indeed, many apparent anomalies in linseed oil drying and a wealth of empirical knowledge in other fields such as internal combustion, the Fischer-Tropsch reaction, and others may fit better into the over-all picture if viewed in the light of the proposed mechanism. For example, the use of organic free radicals as drying promoters can be thought of a8 a simple extension of these views. A difference of opinion exists as to whether the rate determining step is the abstraction of hydrogen or the subsequent polymerization, presumably via free radicals. Regardlees of which view is taken, the proposed electronic explanation of catalysis will apply. The reversible formation of covalent bonds can be pictured as taking place between a 3d electron of the metal and the free radical just as it has been depicted for molecular oxygen.
ANOMALIES I h EXPERIMENTAL DATA
One finding which is not in strict accord with the hypothesis is that among the most active manganese accelerators are included many bidentate amines. The accepted theories of complexing require that chelation occur from coordination with both nitrogen atoms unless structurally prevented from doing so. The fact that the most interesting amines invariably possess structures capable of forming 5- or 6-membered chelate rings implies that two coordinate covalent bonds participate in the structure of the complex. That they perhaps do not enter into combination a t the same time has been proposed ( 4 )as an answer to this objection, particularly in the light of the fact that the only active difunctional amines are characterized by reasonant structures. The first breach in the traditional view that both nitrogens of a bidentate amine react with a metal resulted from the discovery that silver coordinates with two ethylenediamine molecules instead of one, in which case each ethylenediamine may act as a monodentate amine. If this reduction in functionality can occur with as strongly basic an amine as ethylenediamine, one can logically extend the reasoning to the less basic resonating amines
2225
ACKNOWLEDGMENT
This work was supported by Advance Solvents & Chemical Corp. and the Harshaw Chemical Corp., and \vas performed a t the National Printing Ink Research Institute a t Lehigh University, Bethlehem, Pa. LITERATURE CITED
J. S., “Modern Aspects of Inorganic Chemistry,” D. Van Nostrand, N. Y . , p. 11, 1949. Farmer, E. H., and Sutton, D. A , J . Chem. SOC.,1943, p. 126. Gardner, H. A., and Sward, G. G., “Physical and Chemical Examination of Paints, Varnishes, Lacquers and Colors,” 11th ed., p. 157 (1950). Myers, R. R., Ph.D. dissertation, Lehigh University, June 1962. Zettlemoyer, a.C., and Myers, R. R., IND. EFG. CHEX,46, 2220 (1954).
(1) Emeleus, H. J., and Anderson, (2)
(3) (4)
(5)
RECEIVED for review November 3, 1953. ACCEPTED June 8, 1954. Presented before the Division of Paint, Varnish, and Plastics Chemistry a t the 124th Meeting, ACS, Chicago, Ill., September 1953. This paper is from a dissertation submitted by Raymond R. Myers to the graduate schoal of Lehigh University in partial fulfillment of the requirements for the degree of doctor of philosophy, June 1952.