Structures and catalytic reactivities of copper - Wiley Online Library

JOURNAL OF POLYMER SCIENCE: Polymer Chemistry Edition

VOL. 14,1857-1876 (1976)

Mechanisms of Inhibition against the CopperCatalyzed Oxidation of Polyethylene: Structures and Catalytic Reactivities of Copper-Inhibitor Complexes D. L. ALLARA and M. G. CHAN, Bell Laboratories, Murray Hill, New Jersey 07974

Synopsis A series of compounds, mostly derivatives of N,N’-diphenyloxamide, used as inhibitors of copper-catalyzed oxidation in polyolefins has been reacted with cupric salts. The resultant copperinhibitor complexes have been characterized as insoluble, polymeric solids, and their catalytic reactivities in the oxidation of low-density polyethylene were measured at 100°C. The maximum rates of oxidation are first-order in surface area, and some very general trends in structure and reactivity can be drawn from these data. Reasons for the relative effectiveness of various copper deactivators in polyolefin oxidation are briefly discussed.

INTRODUCTION Polyolefins in contact with copper surfaces often undergo accelerated thermal 0xidation.l-5 This presumably occurs through the action of one or more catalytically active copper species, among which copper carboxylate salts have been identified.4*5Such species may be located at both the copper/polyethylene int e r f a ~ eand ~ . ~in the polyethylene matrix.6 Copper inhibitors, often derivatives of oxamide, are added to these systems specifically to retard the catalytic effect of the metal.1-3 Usually other stabilizers (e.g., antioxidants) also are added to increase oxidative stability. However, the presence of the metal deactivator, even in excessive amounts, is usually not sufficient to completely overcome the catalytic effects of the metal. The reasons for this are not clear, and in general, the relationship between the structure of the inhibitor and the magnitude of inhibition is not well understood. In analogy with what has been postulated previously,7 the function of the metal deactivator is to chelate with the catalytically active copper species to form inactive compounds. Hansen and co-workers2y3have reported the formation of a 1:l adduct between N,N’-diphenyloxamide (oxanilide) and copper stearate although the structure was not characterized. In fact, no structures of typical copper-inhibitor compounds have been determined to our knowledge. The actual magnitude of inhibition must be related to the rate at which these chelates form in the polymer, and/or at the metal-polymer interface, and to the residual catalytic activity they possess. While easily defined, these factors are not easily understood. A complication that arises in studying these solid-phase systems, 1857 0 1976 by John Wiley & Sons, Inc.

ALLARA AND CHAN

1858

and often in liquid-phase systems as well, is that various forms of copper, many inhibitors, and many copper-inhibitor compounds are insoluble in the reaction medium. The oxidation and inhibition reactions, therefore, are mostly heterogeneous, and physical as well as chemical factors must be ultimately identified and understood. The scheme (l),although greatly oversimplified, can be used to describe the main features of the mechanism of inhibition: (L),Cu(II)

I

+ I N 2 (IN)Cu(II) + nL

catalytic

k,” oxidation

I

(1)

catalytic

kc’

oxidation

where L is a copper ligand (for example, 02-or RCOz-), IN is an inhibitor molecule, (IN)Cu(II) is the inhibitor-copper complex, and k,‘ and k,N are the global rate constants for the metdl-catalyzed autoxidation of the particular polyolefin medium, In terms of this scheme the effectiveness of an inhibitor molecule is controlled by one or both of the following factors. (1)The catalytically active copper species are scavenged rapidly by the inhibitor molecule with the result that the autoxidation kinetics are controlled by any residual catalytic activity exhibited by the copper-inhibitor complexes; for inhibition to occur we require k,” >> k l . (2) The rate of formation of the copper-inhibitor complex is slow compared to the autoxidation rate (for k:) and controls the effectiveness of the inhibitor along with the residual catalytic activity of any copper-inhibitor complex which may form (still require k,” >> k:): Thus, the rate constant KIN may be controlled by chemical factors and/or physical diffusion and state of reactant contact. There have been numerous studies of the chemical factors in metal ion catalysis, most of these for homogeneous solution.8 However, corresponding studies of the mechanisms for heterogeneously catalyzed autoxidations are very few (when we exclude inappropriately high-temperature and gas-phase systems). Furthermore, the heterogeneous catalysts studied have been restricted mostly to the metal o x i d e ~ . ~Also, J ~ studies of the effects of ligand variation on Cu(II), as well as other metals, are l i m i t ~ d . ~Chalk ? ~ and Smithll have studied the effect of ligand variation in soluble Cu(I1) in cyclohexene autoxidation, but one cannot apply their results directly to the typical multiphase polyolefin systems of interest. As an initial attempt to fill these existing gaps in our understanding of copper inhibition in polyolefin systems, this study was directed toward two objectives. First, a series of inhibitors was reacted with cupric salts to form the types of compounds which are the products of copper ion scavenging in inhibited copper/polyolefin systems. The materials were characterized and attempts made to determine their structures. Second, several chelates of the oxamide family, in the form of insoluble, high surface-area powders, were mixed with polyethylene and their catalytic effects on the oxidation of the polymer studied at 100OC. In addition, several other insoluble copper(I1) compounds with N, 0,and S ligand atoms were examined in order to determine general effects of ligands on catalytic reactivity. In order to compare catalytic rates on a consistent basis, the specific surface areas of the catalysts were measured. This study is intended to provide

COPPER-INHIBITOR COMPLEXES

1859

an understanding of some of the fundamental factors involved in the effectiveness of metal catalysis inhibitors.

EXPERIMENTAL Materials Branched, low-density (0.92) polyethylene was obtained from Union Carbide and contained no inhibitors. The following copper(I1) compounds were obtained commercially and were used without further purification: oxide (Matheson, Coleman, and Bell), sulfide (Fisher), oxalate (K and K Laboratories), glycinate (K and K Laboratories), benzoate (Chemicals Procurement Laboratories), and phthalocyanine (K and K Laboratories). Commercial samples of cuprous oxide (Baker and Adamson) and copper metal powder also were used without additional purification. All the chelating agents were obtained commercially unless otherwise specified.

Preparations The Cu(11)-oxamide compounds were synthesized by reacting the oxamide with cupric acetate [see eq. (2)]. The reactions can be run in any unreactive solvent, and tetrahydrofuran (THF) is a particularly convenient choice. In this solvent, copper carboxylates are soluble, but most of the oxamides and nearly all of the copper chelates are insoluble. Most of the oxamides are partially soluble in dimethylformamide (DMF), and this solvent usually allows reactions to occur faster than in THF. The direction of the reaction can be reversed quantitatively by the addition of strong acid (e.g., aqueous 1M H2S04) to the chelate. This provides a convenient method for analyzing the Cu(I1) content of these compounds since the cupric ion released can be titrated. The reaction product usually contains some unreacted free ligand which can be partially removed by washing with hot DMF. Additional free ligand can be sublimed away by heating in a moderately high vacuum, but the last traces are very difficult to remove. Some chelates such as Cu(0DH) decompose easily on heating (see Table I), and the subliming operation must be done below the decomposition points. In general, it is difficult to purify most of these compounds because of their insoluble and nonvolatile nature. Copper Oxamidate. A sample of 4.0 g of Cu(OAc)a.H20 (20 mmole) was dissolved in 150 ml of refluxing THF (dried over molecular sieve). This solution was added to a stirred slurry of 1.76 g (20 mmole) of oxamide (Matheson, Coleman and Bell) in 50 ml of THF. The mixture was refluxed overnight. The greenish solid was collected by centrifuging the warm mixture. The collected solid was washed once with THF and recentrifuged. The solid was then stirred thoroughly in hot DMF for about 30 min, and the mixture centrifuged while hot. This process was repeated, followed by a final THF wash. The collected solid was dried in uacuo, leaving 2.64 g of a light green powder (88%yield). The powder was purified by heating for 6 hr at 2OOOC and lov5torr pressure. ANAL. Calcd for CuC2H2N202: Cu, 42.5%. Found Cu, 40.4%.

Copper N,N'-Diphenyloxamidate. The preparation for copper oxamidate was followed, except that 3.20 g (13.3 mmole) of N,N'-diphenyloxamide (East-

1860

ALLARA AND CHAN

man) was used in place of the oxamide. After drying in uacuo, 2.84 g green powder was obtained (71%yield). The material analyzed for 18.7%Cu (theory 21.1%). Further purification was obtained by heating overnight a t 29OOC under a vacuum of ton. Some remaining oxanilide appeared to sublime off. Further heating did not remove the last traces of free ligand. ANAL. Calcd for CuC14H10N202: Cu, 21.1%. Found: Cu, 20.2%.

Copper N,N'-Dibenzaloxalyldihydrazidate. The preparation for copper oxamidate was followed, except 4.41 g (15 mmole) of N,N'-dibenzaloxalyldihydrazide (Tennessee Eastman) was used in place of the oxamide, and the mixture was stirred overnight a t room temperature. The final product was pulverized in a mortar and pestle and heated for 24 hr a t 175OC a t a vacuum of torr (until constant weight). A total of 3.08 g of green powder (58%yield) was collected. ANAL. Calcd for CuC16H12N402: Cu, 17.9%. Found: Cu, 17.3%.

Copper Oxalyldihydrazidate. A sample of oxalyldihydrazide (1.18 g, 10 mmole) (Eastman) was dissolved in 150 ml of hot DMF. A concentrated solution of 2.5 g (12.5 mmole) of Cu(OAc)n.H20 in hot DMF was prepared and added to the hot oxalyldihydrazidesolution with stirring. A dark green precipitate formed immediately. The solution was kept hot (below reflux), stirred for 20 min, and filtered hot by suction. The collected solid was stirred in more hot DMF for several minutes and filtered again. This procedure was repeated once more. The solids were washed with T H F and dried in uacuo. After pulverizing, the torr for several hours. The powder darkened solids were heated to 125OC at slightly. A final weight of 1.39 g was obtained (77%yield). ANAL. Calcd for CuC2H4N402: Cu, 35.4%. Found: Cu, 38.2%.

Copper N,N'-Didodecyloxamidate. The preparation for copper oxamidate was followed, except that 1.70 g (4.0 mmole) of the oxamide was dissolved in 75 ml of hot THF. A solution of 0.90 g (4.5 mmole) of Cu(OAc)yH2O in 100 ml of hot THF was added to the first solution and the mixture refluxed under NS for 5 hr. The hot mixture was centrifuged, the supernatant discarded, fresh T H F added to the residue, and the mixture refluxed for about 10 min. The mixture was centrifuged and the washing procedure repeated twice. The final solvent was removed by suction filtration followed by pumping in uacuo. The sample was heated to 15OOC a t torr for 1hr. A green powder was obtained, 6.94 g (48%yield). ANAL. Calcd for CuC26H.&J202: Cu, 13.1%;C, 64.1% H, 10.33%N, 5.75%. Found Cu, 13.0%; C, 63.39%; H, 10.14%;N, 5.92%.

N,N'-Didodecyloxamide. A sample of 1-dodecylamine (7.42 g, 70 mmole) was mixed with 2.92 g (20 mmole) of diethyloxalate and heated to 12OOC for 2 hr. The mixture was cooled and the product recrystallized from hot benzene (7.35 g, 87% yield). ANAL. Calcd for CzsHszNzOz: C, 73.4% H, 12.3% N, 6.58%. Found C, 73.28%; H, 12.32%;N, 6.49%.

Copper N,N'-Bis( o-nitropheny1)Oxamidate. A solution of 1.22 g (6.1 mmole) of Cu(0Ac)yHzO in 100 ml of DMF was added to a stirred slurry of 0.99


1861

g of the oxamide in 100ml of DMF. After stirring for 30 min, the solution became clear with deep green color. The solution was stored overnight and then poured into 400 ml of water. The flocculent green precipitate was filtered by suction, washed three times with water, once with THF, and dried in uacuo. The final product (0.99 g, 84% yield) is soluble in DMF and dimethyl sulfoxide, slightly soluble in a pyridine-methanol mixture (5:1), and insoluble in THF diglyme, acetonitrile, and ethanol. ANAL. Calcd for C U C ~ ~ H ~ NCu, ~ O16.23%. ~: Found Cu, 10.2%.

N,N'-Bis( o-nitropheny1)Oxamide. A solution of 5.52 g (40 mmoles) of o-nitroaniline in 70 ml of THF was stirred, while a solution of 1.26 g (10 mmole) of oxalylchloride in 7 ml of THF was slowly added from a dropping funnel. An immediate yellow precipitate was formed. The solution was stirred an additional 5 min, then poured in 300 ml of 1M HC1, and the solid filtered by suction. The product was washed several times with 1M HC1, several times with water, and finally with THF until the filtrate was colorless. The yellow powder was dried in uucuo (2.71 g, 82% yield). ANAL. Calcd for CI4H10N406: C, 50.9%; H, 3.05%; N, 16.95%. Found: C, 50.22%; H, 3.09%; N, 16.67%.

Copper [N,N'-Bis( o-hydroxybenza1)oxalyl Dihydrazidate] (1:l). A sample of 0.62 g (1.9 mmole) of the d i h y d r a ~ i dwas e ~ ~dissolved in 50 ml of DMF. A solution of 0.42 g (2.1 mmole) of Cu(OAc)a-HaOin 50 ml of DMF was added. The solution immediately turned a deep, dark yellow color but appeared clear. After stirring for 15 min, the mixture was poured in 200 ml of water and the greenish slurry filtered very slowly by suction, leaving a pasty green solid which was washed once with water, refiltered, and dried in uucuo (0.62 g, 84%). The green-yellow powder was insoluble in most solvents except hot DMF. ANAL. Calcd for CuC16H12N404: Cu, 16.4%. Found: Cu, 14.8%. Calcd for C U C ~ ~ H ~ ~ N ~ O K 2H20 CU,15.0%.

Copper [N,N'-Bis( o-hydroxybenza1)oxalyl Dihydrazidate] (2:l). The above procedure for the monocompound was followed, except the quantity of copper acetate was doubled. The reaction mixture was allowed to stand overnight. The dark green slurry was poured into 500 ml of water, the mixture centrifuged, and the supernatant discarded. The dark green sludge was stirred in 75 ml of hot DMF for 15min and centrifuged. This process was repeated two times. The sludge was mixed with the THF, centrifuged, collected, and dried in uacuo. On the basis of two coppers per molecule, a 91% yield was obtained. ANAL. Calcd for C U & ~ ~ H ~ , J NCu, ~ O28.3%. ~: Found Cu, 24.7%.

Copper [N,N'-Bis(3-dimethylamino-l-propyl)Oxamidate]. A solution of 1.03 g (4.0 mmole) of the oxamide in 50 ml of benzene was mixed with 0.80 g (4.0 mmole) of Cu(OAc)z-HzO,and the mixture was refluxed until all the copper acetate dissolved, at which point the solution was a dark, inky blue. The solution was concentrated to about 10 ml. Upon cooling to about 18"C, a precipitate formed. The blue solid was collected by suction filtration and dried in uucuo (0.64 g, 50% yield). ANAL. Calcd for CuC12&&4' 02:

Cu, 19.9%. Found Cu, 22.9%.

1862

ALLARA AND CHAN

N,N'-Bis(3-dimethylamino-l-propyl)Oxamide. 3-Dimethylamino-lpropylamine (5.11 g, 50 mmole) was mixed with 3.7 g (25.3 mmole) of diethyl oxalate, and the mixture heated a t 12OOC for 2 hr. After cooling to room temperature, the solid product was recrystallized from pentane to give white crystals. ANAL. Calcd for C12H26N402: C, 55.7%; H, 10.12%;N, 21.8%. Found: C, 55.85% H, 9.90%;N, 2 1.48%.

Copper Salicylalethylenediimine. A solution of 2.6 g (10 mmole) of a,a'(ethylenedinitri1o)di-o-cresol(Eastman) was dissolved in 60 ml of tetrahydrofuran. To this was added, with stirring, a solution of 2.0 g (10 mmoles) of Cu(0Ac)yHaO in 100 ml of tetrahydrofuran. The mixture was stirred about 15 min and the green precipitate collected by suction filtration. The precipitate was washed with tetrahydrofuran until the filtrate was colorless and then dried in uucuo. The final product weighed 3.10 g (94%yield). The analysis indicated one water of hydration. ANAL. Calcd for C16HIoNzOz: Cu, 19.3%. Calcd for C & I ~ N ~ O ~ - HCu, ~ O18.3%. Found Cu, 18.2%.

Copper Hydroxybenzoate. Cuprous benzoate was first prepared by modifying the method of Cohen and L e ~ i n .Benzoic ~ ~ acid (0.11 mole) was dissolved in 200 ml of xylene. Cuprous oxide powder (0.025 mole) was added and the mixture refluxed under nitrogen overnight with a Dean-Stark trap connected to remove water. After reaction only a trace of red oxide remained, and after cooling in ice a white solid formed. The xylene was filtered off under nitrogen pressure and the solid washed with 50 ml of xylene and 25 ml of ether. The solid was then freed of remaining solvent by pumping a t 0.2 torr, leaving 8.5 g of a whitish-gray solid (42%~ yield). The material discolors in air, and if it is necessary to store it, the sample must be placed in a light-proof, well-stopped bottle. ANAL. Calcd for CuC~H602: Cu, 34.4%. Found: Cu, 34.1%.

A portion of the cuprous benzoate sample was pulverized in a mortar and pestle and allowed to stand in contact with air for several days. The sample turned blue. ANAL. Calcd for C U C ~ H ~ O Cu, ~ :31.6%. Found Cu, 32.0%.

The spectrum (Nujol mull) resembles those recorded for metal b e n z o a t e ~ ~ ~ with major absorptions a t 1594 cm-' (C-C), 1551 cm-l (antisymmetric C=O stretch), and 1412 cm-l (symmetric C=O stretch). Additional absorptions in the hydroxy benzoate spectrum appear a t 3604 cm-l(O-H stretch) and 910-940 cm-l (probably M-OH bending modes).37

Kinetic Procedures Samples for kinetic runs were prepared by the mixing of weighed amounts of catalyst and low-density polyethylene on a two-roll mill heated to about 130OC. The total mixing time for each sample was about 5 min. After mixing, several grams of the milled sample were pressed into 18-22 mil films between clean Mylar sheets with the temperature controlled to 110-120OC. The contact time in this temperature range was less than 30 sec. The films were cut into 0.5-cm wide strips weighing about 100 mg and placed into a standard manometric apparatus


1863

for measuring the consumption of oxygen. All measurements were made a t lOO(*l)"C, under 1 atm of pure oxygen. The volume of oxygen consumed is related in nearly exactly the same way for each run to the number of moles of oxygen consumed, since the total reaction volumes are the same within several percent from run to run. Therefore, plots of oxygen uptake are presented most simply by volume plots. All runs were made with at least duplicate samples and, in some cases, with quadruplicate samples. The initial specific surface areas of the copper compound used were determined, and where possible, the compositions of the polymer-catalyst mixtures were adjusted to cover similar ranges in the ratio of catalyst surface area to weight of polymer. Differential thermal analyses showed that all of the copper compounds were stable toward melting or decomposition a t the milling, molding, and reaction temperatures. None of the compounds exhibited any visual signs of solubility in hexadecane at 15OOC for a 0.01% (by wt) composition mixture.

Other Measurements Surface area measurements were carried out on a Perkin-Elmer sorptometer using the BET method. All samples were dried a t 75-1OOOC under a vacuum torr prior to surface area measurement. Thermal stabilities were meaof sured on aDuPont 950 thermogravimetric analyzer. Infrared spectra were recorded on a Perkin-Elmer 621 spectrophotometer using Nujol-mull methods. The technique was used with NaCl plates in the 4000-600 cm-l region and AgCl plates in the 600-300 cm-' region. The magnetic susceptibilities were measured by using a pendulum magnetometer between 1.5 and 300°K a t an applied field of 15,300 Oe.

RESULTS AND DISCUSSION Structures and Properties of Copper-Deactivator Compounds Solubilities and Stoichiometries. The compounds of Cu(I1) and oxamide derivatives were prepared by metathesis reactions of cupric acetate with the particular oxamide [eq. (2)], 0

Cu(OAc),

+

0

II

(RNHC%

1I

--+

Cu[(RNC+]

+ 2HOAc

(2)

where R may be H (oxamide), CsH5 (oxanilide), CGH~CH=N- (N,N'-dibenzaloxalyl dihydrazide, BODH), n-C12H25-, NHz-(oxalyl dihydrazide, ODH), and other similar groups. Scavenging of cupric salts in a polyethylene matrix should occur in a similar manner. The analysis and the corresponding stoichiometry for each of the compounds are given in Table I. Most of the compounds are quite insoluble in both dimethylformamide (DMF) and tetrahydrofuran (THF), as well as a variety of other polar and nonpolar solvents. This observation-together with the analyses, infrared spectra, and magnetic properties (see below)-are best explained on the basis of polymeric structures of the type I. A similar type of structure has been proposed for the copper complexes of N,N'- disubstituted dithiooxamides.12 These structures require one extra ligand molecule per chain as the end group, and this accounts for the observed tenacity of the last traces of ligand. The number of repeat units for these proposed

16.2 16.4g 28.3” 19.9

10.2 14.8 24.7 22.9

-

-

13.0d

S-hotli S-hot/i

Large Large

e

-

e e

s/s

-1-

i/i S-hot/i

i/i

S-hotli

26

S/S

S/i S/i i/i

i/i

i/i

i/i

S-hotli

19

i/i

Cu compound

S-hot/i

24

Free ligand

Solubility (DMF/THF)b

Light green Yellow-green Dark green Blue

Green

Dark browngreen Brown-green

Light green

Light green

Color

-

-

-

38

385

-

6-23

28

40

34

E,, kcal/ mole

125-220

295

380

315

t,%, “C

Thermal decompositionc

Reflux, 16 hr Reflux, 16 hr Ambient, 16 hr Ambient, 1 6 hr Reflux, 5 hr f f f f

Conditions

48

14

85

80

88

%

Yield,

Ease of formation in THF

a Estimated from analysis data assuming that the discrepancy between the observed and calculated copper analysis values is due to the contribution of end groups. b Solubility: S = soluble; i = insoluble. C Thermogravimetric analysis. d Also consistent with C, H, N analyses. e Structure assigned as monomeric based on other evidence (see text). f Reactions not run in THF. g Cu(OX).2H2O;CU, 15.0%d c d . h Value calculated for two copper atoms per molecule.

(o-NOZ)C6H5(0-OH)C6H5CH=N(0-OH)C6H5CH=N(CH, )”H, 13-

13.0

13.1

35.4

38.2

35.4

NH,-

17.3

17.3

17.9

20.2d

20.2

21.1

40.5

40.4

Found

Calcd for Value of n in [(OX) Cu],. [(OX)Cu], OXH, (OXH,),

42.5

Calcd for 1 : l

C6H5CH=N-

c6H5-

H-

2

Ligand group R

cu, %

TABLE I Chemical and Physical Properties of Synthesized Cu( 11)-Oxamide Compounds

z

F

0

i2


1865

polymeric structures can be estimated from the analyses (Table I) and range from ca. 20 to higher values.

I

The alternative structure (11)is also possible and is consistent with all our data (see below also) except that, in contrast to structure I, a higher solubility might be expected, since a solvent molecule (e.g., DMF) would not need to rupture a Cu-N bond to release a solvated monomer group.

I1

An x-ray analysis of the powdered materials shows that the structures of the oxamide and oxanilide copper compounds are highly disordered in contrast to the free ligands, which are highly ordered crystals. Apparently, the polymeric chains maintain almost no ordering with respect to one another. The presence of additional coordinating groups on the oxamide can interrupt the polymeric structure. Ojima and Yamada13 have prepared monomeric complexes such as the bis-N,N'- (3-aminopropy1)oxamide complex (111). Similarly, our observations (Table I) of high solubilities indicate that the ligands

I11

o-nitrooxanilide and N,N'- bis(3-N,N-dimethylamino-l-propyl) oxamide form monomeric complexes with structures similar to I11for the latter and speculated as shown in IV for the former. The Ni(I1) complex with the thiooxamide corresponding to the latter ligand is reported

N-0' 0

'0-N 0

1866

ALLARA AND CHAN

to be insoluble in water and ethanol and is presumed to have a polymeric structure.12 The ligand N,N'- bis(o -hydroxybenzal) oxalyl dihydrazide (0-0HBODH), which is capable of chelation at two different sites, forms a soluble complex with the first mole of copper by binding at the phenolic groups (V), as shown by infrared evidence (see below). The second mole of copper binds at the amide nitrogens to form an insoluble complex which is most likely polymeric.

V

Infrared Spectra. Complexation causes noticeable changes in the NH and C=O stretching vibrations of the oxamide derivatives. These absorptions are summarized in Table I1 and compared with published14spectra of oxamide and the bisoxamido complex (VI).

VI

Oxamide itself has two N-H stretching bands, 3280 and 3170 cm-l. These are replaced upon chelation with a single band a t 3250 cm-l [Cu(II) oxamidate] or 3310 cm-l (bisoxamido complex). This change in the N-H absorption has been attributed14 to the replacement of an N-H bond on the free oxamide by an N-Cu bond in the complex. The oxamide derivatives appear to complex in a similar fashion, since the amide N-H stretching frequency generally is lost upon formation of the Cu(I1) complex. The N-H band remaining in the dihydrazide complex is presumably the amine N-H, while the remaining N-H band in the mono-o-OH-BODH complex is consistent with the proposal that the first copper is coordinated through the OH substituent.* The 3485 cm-' absorption found in the complex of the dimethylamine-1-propyl derivative is high compared to the other NH absorptions. However, the compound is not sufficiently pure (Table I) to exclude water or some other impurity as the source of this latter absorption. Considerable variation in carbonyl absorption is seen upon complexation. In those cases where assignments can be made, the C=O stretching frequency of the free ligands shifts by 27-125 cm-l to lower frequencies in the complexes. A * The OH stretching frequency is not found in either the DMF-recrystallized o-OH-BODH or the copper complex. In the free ligand, the band is shifted to lower frequencies due to hydrogen bonding with the C=N group and is obscured by the C-H bands. A model compound, a,a'-(ethylenedinitri1o)di-o-cresol,shows very similar behavior.

3280 (b), 3170 (b) 3280, 3200 (sh) 3294 3308 3300 3244 3251e 3280 (b), 3170 (b)

Free ligand

3250f 3310

-

3250 3300-3200 (w)b 3485

Cu(I1) complex

Cu(I1) complex 1590 (b) 1650-1610 (b)’ 1622 1595 (b) 1554d 1550 g 1620, 1590C9”

Free ligand 1655 (b) 1652,1648C 1649 1644 1662 1662,1652C 1658e 1655 (b)

b

a

(b) = broad, (sh) = shoulder, (w) = weak. Poorly resolved spectrum. C Band is split. d 1560 cm-’ was reported by Hansen e t aL3 for the reaction product (in polyethylene) of copper stearate and oxanilide. e Recrystallized from DMF. f Monocomplex. g The C=O absorption is not assigned because of the complexity of the spectra in the region. h Data of Armendarez and Nakamoto;14 the complex in this case is K,~CU[(NHC=O)~],.

C, H CH=N(o-OH)C6H3CH=NH--, bis complex

c6H5-

HNH,(CH3)2N(CH2)3-c ,*HZ5-

Ligand group R

TABLE I1 Infrared Absorptions of Oxamide Compounds and Their Cu(1I) Complexes

-( 35/65)

-65 5 X k, 850 ml/hr-m*. a

b

*lo%. No similar studies have been carried out in the solid phase, but Mukherjee and Graydon9 have followed the kinetics of the liquid-phase oxidation of tetralin over CuzO. They observe a slightly lower than first-order dependence of rate on the amount of catalyst in contrast to our first-order behavior in eq. (3). Their rate law has the form:

where k is the empirical rate constant, [RH] is the tetralin concentration, and M is the catalyst mass per unit volume of hydrocarbon. However, our data for CuzO do not fit eq. (3) over the whole range of values (Fig. 2). The slope decreases (not smoothly) from low to high surface areas and thus gives a lower than first-order behavior.

-"

" I

10

5 u,mz/g

x

103

Fig. 5. Ro vs. u for Cu(OH)(OBz),Cu(OBz)z,and Cu(oxa1ate).

ALLARA AND CHAN

1872

0 O x :

PHTHALOCYANINE

OXALATE

OXAMIDE

O 0 ' x : :

I

N= CH-CgHg BODH

GLYCINATE

I CgH5 OXANlLlDE

HYDROXY, BENZOATE

BENZOATE

SALICYLALETHYLENEDlMlNE

Fig. 6. Structural representations of organic ligand groups.

Some trends of catalytic activity as a function of structure can be drawn from the data in Table 111. For convenience, the structures of the organic ligands and the relative reactivities of their corresponding copper complexes (taken from Table 111)are given in Figure 6 in order of increasing reactivity. The location of the negative charges are shown for each ligand (neglecting obvious delocalization). All complexes with negatively charged nitrogen ligands give low reactivities. Copper bound only to oxygen atoms of unit negative charge (benzoate and hydroxy benzoate) shows moderate reactivity except for oxalate, which gives unusually low reactivity. If copper bound to negatively charged oxygen is additionally coordinated by neutral nitrogen (glycinate, salicylalethylenediimine), the reactivity is still moderate. When sulfide or oxide dianions are associated one-to-one with copper (CuO, CuS), the reactivity is low. The highest reactivities are shown by cuprous oxide and a copper surface exposed to air a t ambient temperatures. The latter surface is reported to be cuprous oxide,lgand may exhibit a gross defect structure of stoichiometry CuO0.67.~~ When temperatures exceed 2OOOC a layer of CuO begins to form.21 Thus, observed differences in catalytic activity between oxidized copper and stoichiometric CuzO are reasonable. Catalytic Mechanisms. In order to understand the relationship between structure and reactivity, it is necessary to know the actual chemical reactions involved in the catalysis. In most cases, the kinetics of free-radical-initiated autoxidations of saturated hydrocarbons are determined by the sequence of reactions comprising initiation, propagation, and termination steps. The addition of a soluble metal catalyst to the oxidation system generally accelerates the initiation rate, rarely has an effect on propagation reactions, and in a few cases may affect termination rates.8 Much less information is available for heterogeneously dispersed, solid catalysts; but they appear to behave similarly in many


1873

respects, particularly in regard to the formation of reaction products.1° In general, it appears from other work that the types of initiation processes which most likely occur in our oxidations involve the decomposition of hydroperoxides to free radicals. For these reactions, the metal can be considered to act as a redox center, either accepting or donating an electron. Hart and Ross23have studied the gas-phase decomposition of tert- butyl hydroperoxide over oxidized copper surfaces. They propose an electron-transfer reaction as an initiation step. In their study of the oxidation of tetralin over Cu20, Mukherjee and Graydong proposed a surface decomposition of hydroperoxide to alkoxyl and hydroxyl radicals. In recent studies with cobalt acetylacetonate, Stivala, Jadrnicek, and Reich22have observed that the maximum rates of oxidation for atactic polypropylene (bulk phase) under constant oxygen pressure at 110°C obey rate laws similar to eq. (3). These workers postulate hydroperoxide decomposition reactions to explain the function of the metal catalyst in radical initiation. In addition, these authors suggest a direct reaction between polypropylene, 0 2 , and the metal catalyst. We consider the latter proposal unlikely, particularly in polyethylene, because of the high C-H bond strengths and ionization potentials of the hydrocarbon -CH2groups. It follows that the major catalytic reactions in the present study should be surface catalyzed decompositions of hydroperoxides. However, this conclusion is probably inappropriate for air-aged copper and cuprous oxide at the long reaction times of the maximum rate stage of the oxidation, where our previous ~ o r kindicates ~ , ~ that the copper oxide surface should be completely converted to a copper carboxylate phase. The major catalytic species would then be cupric ions, at the interface and diffused into the matrix.6 The observed differences between the reactivity of Cu20 and CuO are quite significant (Table I11 and Hansen et al.l) but not simply explained. Variation in reactivities between the two oxides have been attributed to the differences in semiconductor properties24 and the types of copper-oxygen surface species which may be present.25 However, it is obvious that these factors must be incorporated into both the rates of hydroperoxide decomposition and carboxylate forming reactions at the surface. Excluding the oxides, it seems reasonable to describe the catalytic reactions on a basis in which only the properties of the reacting metal ion and its surrounding ligands at the surface are considered. Such descriptions have been used for the mechanisms of chemisorption and catalysis over transition metal catalysts.2628 In support of this approach, Gould and RadolO have found close parallels between the productk of cyclohexene autoxidation with heterogeneous and homogeneous catalysts of the same transition metal. One of the properties of transition metal ions which is commonly invoked to explain differences in catalytic reactivity is the redox potential.* An apparent correlation has been made recently between the redox potentials (presumably for aqueous ions) and the catalytic activities of a series of transition metal acetylacetonates (highly dispersed) in the bulk-phase autoxidation of polypropylene at l10°C.29 However, the effects of changing the ligands on a copper(I1) ion in the homogeneous solution autoxidation of cyclohexenell do not appear to follow any one property, such as redox potential, but seem to involve additional properties, such as the ability to coordinate hydroperoxide molecules. (In fact, examples of catalysis are known for copper where the catalytic mechanism of a hydrocarbon oxidation

1874

ALLARA AND CHAN

may change completely with a change of ligands on copper.30) The values of k , in Table I11 appear to be influenced somewhat by the same factors that influence redox potentials. The strongly basic, negatively charged amido ligands in the oxamide family and phthalocyanine are expected preferentially to stabilize the Cu(1I) state relative to Cu(1) or Cu(0). Further, the bidentate and tetradentate coordination will additionally stabilize Cu(I1). For example, James and William~ observe ~ ~ that bidentate amine ligands are more effective a t stabilizing Cu(I1) than the corresponding monodentate ligands. As the stability of Cu(I1) increases relative to Cu(I), the redox potential changes, and Cu(I1) becomes a less powerful oxidizing agent and presumably a less active oxidation catalyst. As the ligands are changed to the less basic, singly charged oxygen anions (carboxylate and hydroxide), capable of only monodentate binding, the ability to selectively stabilize the Cu(I1) state should decrease, and the Cu(I1) thus becomes a better oxidizing agent. The low value of k , for oxalate appears to be due to the additional stability of Cu(I1) imparted by bidentate binding. In sharp contrast to our results, copper oxalate has been suggested as an active catalyst for the liquid-phase oxidation of paraffins over copper surfaces.32 Additional binding by neutral nitrogen ligands (glycinate and salicylalethylenediimine) should increase stabilization of the Cu(I1) state, perhaps primarily because of increased c ~ o r d i n a t i o nand , ~ ~ decrease the catalytic activity. However, little change is observed. In regard to the comparison between the inorganic ligands in CuS and CuO, one might expect the more polarizable and less basic S2- ion to stabilize Cu(I1) relative to Cu(1) less effectively than 02-. On this basis, CuS should be more active than CuO as is observed (Table 111). Kinetic Effects in Inhibition. The reasons for the different performance of copper inhibitors in polyolefin compositions are not known usually. The present data on the residual catalytic activity of copper-deactivator complexes should be useful in analyzing relative deactivator performance data. It is unfortunate that few published data exist on the oxidation kinetics of carefully controlled polyolefin composites containing copper deactivators. However, some interesting comparisons can be made using limited data available. For example, Hansen and co-workers1 reported that typical chelating agents such as N,N’disalicylidene-1,2-propanediamineare virtually ineffective in copper-antioxidant-polypropylene composites, whereas compounds of the oxamide family are quite effective. This result is reasonable on the basis that the kinetics of chelate formation are very fast (i.e., not rate-determining), and the differences are due to the differences in reactivities of the copper chelates. The former compound also, our above is very similar in structure to a,a’-(ethylenedinitri1o)di-o-cresol; data show that the copper chelate of this structure (salicylalethylenediimine) is much more reactive than the copper-oxamide chelates (Table I). On the other hand, a comparison of polyethylene-copper samples33 containing a typical phenolic antioxidant shows that the oxidation induction period at 12OOC with added N,N’-dibenzaloxalyldihydrazide(0.1%)is 2.8 times as great as for added oxanilide (0.1%). Since our data show that the copper chelates of both these inhibitors are of equal reactivity, it suggests that the differences between these inhibitors under these conditions are due to the rates at which the active copper species are scavenged. This rate may be related both to the chemical structure and the state of dispersion and persistence of the inhibitor. We have outlined above the chemical and physical factors which can control


1875

the effectiveness of a metal deactivator. The present results show that deactivators can form copper compounds with residual catalytic activity. General relationships between deactivator structures and the corresponding structures of the copper compounds have been established. Further, structure-reactivity trends of these copper compounds for polyethylene oxidation have been determined. Such trends could be useful in selecting new deactivators. However, a t present it is clear that other types of chemical and physical factors may often control inhibitor effectiveness. These factors include such complex phenomena as inhibitor dispersion, solubility, and persistence, which are difficult to evaluate. It is obvious that evaluation of such effects, together with the present study, could provide a significant advance in the fundamental understanding of metal deactivator effectiveness. The authors wish to acknowledge the cooperation of R. C. Sherwood and F. Schrey for obtaining magnetic and surface area measurements.

References 1. R. H. Hansen, C. A. Russell, T. DeBenedictis, W. M. Martin, and J. V. Pascale, J. Polym. Sci. A, 2,587 (1964). 2. R. H. Hansen, T. DeBenedictis, and W . M. Martin, Polym. Eng. Sci., 5,223 (1965). 3. R. H. Hansen, T. DeBenedictis, and W. M. Martin, Trans. Znst. Rubber Ind., 39,290 (1963). 4. M. G. Chan and D. L. Allara, J. Colloid Interface Sci., 47.697 (1974). 5. M. G. Chan and D. L. Allara, Polym. Eng. Sci., 14,12 (1974). 6. D. L. Allara, C. W. White, R. L. Meek, and T. H. Briggs, J.Polym. Sci. Polym. Chem. Ed., 14, 93 (1976). 7. G. Scott, Atmospheric OXidQtiOn and Antioxidants, Elsevier, New York, 1965. 8. L. Reich and S. S. Stivala, Autoxidation of Hydrocarbons and Polyolefins, Dekker, New York, 1969, Chap. 4. 9. A. Mukherjee and W. F. Graydon, J. Phys. Chem., 71,4232 (1967). 10. E. S. Gould and M. Rado, J. Catal., 13,238 (1969). 11. A. J. Chalk and J. F. Smith, Trans. Faraday SOC.,53,1235 (1957). 12. R. N. Hurd, G. DeLaMater, G. C. McElheny, and J. P. McDermott, Advances in the Chemistry of Coordination Compounds, S. Kirschner, Ed., Macmillan, New York, 1961, p. 350. 13. H. Ojima and K. Yamada, Nippon Kagaku Zasshi, 89,490 (1968); Chem. Abstr. 69,40822a (1968). 14. P. X. Armendarez and K. Nakamoto, Znorg. Chem., 5,796 (1966). 15. B. N. Figgis and J. Lewis in Modern Coordination Chemistry, J. Lewis and R. G. Wilkins, Eds.. Interscience, New York, 1960, Chap. 6. 16. M. Kato, H. B. Jonassen, and J. C. Fanning, Chem. Reu., 64,99 (1964). 17. R. N. Hurd, G. DeLaMater, G. C. McElheny, and L. V. Peiffer, J. Amer. Chem. Soc., 82,4454 (1960). 18. H. Ojima and K. Nonoyama, 2. Anorg. Allg. Chem., 389,75 (1972), and papers cited therein. 19. J. Kruger, J. Electrochem. SOC.,106,847 (1959). G. Poling, J. Electrochem. Soc., 116.958 (1969). 20. H. Wieder and A. W. Czanderna, J. Phys. Chem., 66,816 (1962). 21. G. Poling, J. Electrochem. Soc., 116,958 (1969). 22. S. S. Stivala, B. R. Jadrnicek, and L. Reich, Macromolecules, 4,61 (1971). 23. A. B. Hart and R. A. Ross, J. Cat& 2,121 (1963). 24. 0. V. Krylov, Catalysis by Nonmetals, Academic Press, New York, 1970, Chap. 1. 25. D. L. Allara and R. F. Roberts, to be published. 26. D. A. Dowden and D. Wells, Actes 2e Cong. Intern. Catalyse (Technip, Paris), 2, 1489 (1961). 27. J. L. Garnet and W. A. Sullich-Baumgarten, Adu. Catal., 16,95 (1966). 28. J. J. Rooney and G. Webb, J. Catal., 3,488 (1964). 29. L. Reich, B. R. Jadrnicek, and S. S. Stivala, J. Polym. Sci. A-1, 9,231 (1971).

1876

ALLARA AND CHAN

D. L. Allara, J . Org. Chem., 37,2448 (1972). B. R. James and R. J. P. Williams, J . Chem. SOC.,1961,2007. G. W. Poling, J . Electrochem. SOC.,117,520 (1970). J. J. Mottine, unpublished results. R. L. Hartless and A. M. Trozzolo, Organic Coatings Plastic Preprints, 34(2), 177 (1974). T. Cohen and A. H. Lewin, J . Amer. Chem. SOC.,88,4521 (1966). J. H. S. Green, W. Kynaston, and A. S. Lindsey, Spectrochim. Acta, 17,486 (1961). K. Nakamoto and P. J. McCarthy, Spectroscopy and Structure of Metal Chelate Compounds, Wiley, New York, 1968. 30. 31. 32. 33. 34. 35. 36. 37.

Received September 8,1975 Revised October 24,1975