April 20, 1957
THERMAL DECOMPOSITION O F DI-t-BUTYL [CONTRIBUTION FROM
THE
NITRATEESTERS
P E R O X I D E AND
SCHOOL OF CHEMISTRY, USIVERSITY
OF
1703
MINNESOTA]
Infrared Investigation of the Thermal Decomposition of Di-f-butyl Peroxide and Nitrate Esters in Potassium Halide Pressed Disks, and the Spectrum of Carbon Dioxide at High Pressures1 BY HENRYA. BENTAND BRYCECRAWFORD, JR. RECEIVEDDECEMBER 7 , 1956 Dispersion in potassium halide pellets appears to be a safe and convenient way to study spectroscopically decompositioii of unstable compounds. In the infrared, the thermal decomposition of di-t-butyl peroxide has been followed in pellets of KBr up t o relatively high temperatures, products identified, and the rate constant for scission of the 0-0 bond found to be 1.45 X 1016e-38.0(k”1.)lRTsee.-‘. Nitrate esters produce NOa- and COZin KCl, KBr and KI. Some temperature dependent oxygen atom free energy levels are developed for the anhydrous nitrate-potassium halide system, and spectra are given of COe under apparently high compression.
During an investigation of the flame decomposition of a nitroglycerine-nitrocellulose double-base propellant by infrared spectroscopy, we obtained a nice spectrum of nitrocellulose dispersed in a potassium bromide matrix, following the pellet technique recently developed by Stimson and O’Donnel for obtaining the spectrum of solid^.^,^ On heating, new bands appeared in the spectrum of these pellets. It occurred to us that this might, in fact, be a convenient, and safe, way to follow the thermal decomposition of nitrocellulose; as well, perhaps, as other unstable compounds. The initial step in the decomposition of nitrate esters, most investigators agree, is scission of the 0-NO2 bond to give NO2 and an alkoxyl radica14-6
ROn’Oz +RO. $. KO2
(1)
What happens after this is less certain. The NO2 complicates matters not just a little. Simpler, but similar in the initial production of alkoxyl radicals, is the decomposition of organic alkyl peroxides, the first step being scission of the RO-OR bond to give two alkoxyl radicals RO-OR
+2RO.
(2)
followed by radical dismutation, recombination and hydrogen abstracti~n.’?~ We chose to look first a t di-t-butyl peroxide. Pioneering work on the decomposition of this peroxide has been carried out in several laborat o r i e ~ . $ - ~At ~ room temperature it is a liquid, and fairly stable. Experimental Pellet Preparation.-Pellets were manufactured on a Carver hydraulic press (20,000 lb. thrust on a 1.25 in. ram) (1) This research was supported by the U. S.Navy, Bureau of Ordnance, through contract with the University of Minnesota. 74, (2) (a) M . M. Stimson and M. J . O’Donnel, THIS JOURNAL. 1805 (1952); (h) U . Schiedt and H. 2. Rheinwein, Nalurforsch., 76, 270 (1952). (3) J. B. Jensen, Acta Chem. Scand., 8 , 393 (1954). (4) G. K. Adams and C. E. H. Bawn, Trans. Faraday Sor., 46, 494 (1949). (5) L. Phillips, N a f u r e , 165, 564 (1950). (6) J . B . Levy, THISJOURNAL, 76, 3254, 3790 (1954). (7) A. V. Tobolsky and R . B . Mesrobian, “Organic Peroxides,” Interscience Publishers, Inc., New York, N . Y., 1954, pp. 57-72. (8) E. W. R . Steacie, “Atomic and Free Radical Reactions,” Reinhold Publ. Corp., New York, N. Y., 1954, pp. 232-236. 68, 205, 643 (9) N . A. Milas and D. M. Surgenor. THISJOURNAL, (1946). (10) P. George and A. D. Walsh, Trans. Faraday SOL.,42,94 (1946). (11) J. H. Raley, F. F . Rust and W. E. Vaughan. THIS JOURNAL, TO, 88, 1336 (1948).
in a 2 in. X 1.25 in. die, which could be evacuated, equipped with plungers of hardened steel 0.5 in. in diameter. The potassium halides-KC1, KBr and KI-were well ground, oven dried and reground before using. Other inorganic salts, such as potassium nitrate, were dry-mixed with the matrix, sometimes with a little Wig-L-Bug vibrator, but usually by hand. Liquids-such as ethyl nitrate and the di-t-butyl peroxide (obtained from Jasonols Chemical Corp ., Brooklyn, N. Y.)-were added dropwise from a syringe t o the powdered potassium halide resting on the bottom plunger in the die, just prior t o pressing. Pellets with liquids dispersed in them-we have tried also acetone, methyl ethyl ketone, chloroform and a-chloronaphthaleneare usually cloudy; this, however, does not seem t o bother particularly in the infrared Our optically best nitrocellulose dispersions were obtained by saturating finely powdered KCl, KBr or K I with a non-aqueous solution of the nitrate ester, followed by slow evaporation of the solvent, careful mixing and regrinding with roughly an equal amount of fresh potassium halide, and compression. Our pellets were 0.5 in. in diameter and 1-15 mm. thick. Heat Treatment.-In preliminary, and later vacuum-line experiments, pellets were heated under an infrared lamp; in kinetic runs, in a thermostated oven, temperature constant to zkO.25” or better. Invariably pellets become increasingly cloudy on heating and sometimes even on standing a t room temperature. R’e suspect devitrification of the matrix, a process that might be expected t o accelerate with rising temperature; and we were not able to stop this in several trials with mixed matrices, like KC1-KBr; indeed, this only seemed t o intensify the problem. Consequent scattering and deterioration of optical transparency was often severe above 2000 cm.-l. Recompression.-This technique proved quite useful in cases of severe scattering. Devitrified pellets were simply put back in the die and recompressed. Often one-quarter to one-tenth the initial compression pressure was sufficient t o restore optical transparency. Spectra.-Surveys and kinetic data were run on a PerkinElmer Model 21 double-beam spectrometer equipped with sodium chloride optics; that of COZ trapped in potassium halide matrices on a model 12C double-pass spectrometer with a lithil In fluoride prism. For low temperature liquid nitrogen rum pellets were mounted in a cold cell similar t o that described by Wagner and Hornig.12 I
Results and Discussion Thermal Decomposition of Di-t-butyl Peroxide in KBr.-The radical dismutation (CH,)BCO.
+
( C H I ) ~ C O CH,
hydrogen abstraction CH3. -+- CHa and (CHJ)3C0.-+ (CH1V.20H
and recombination .CH*(CH,),CO.
---+ (CH3)2C--CH2
Y
(12) E. L. Wagner and D. I? Hornig, J . C h c m P h y s , 18, 299 (1950).
HENRYA. BENTAND BRYCECRAWFORD, JR.
17!M
VOl. 79
100
-
C
. I
v1
.-,.
2
G
5o
s 0 3500
3000
2500
2000 1900 1800
1700 1600 1500 1400 1300 1200 1100 1000 900 800 700 Frequency (cm - l ) . Fig. 1.--Spectra of di-t-butyl peroxide arid decomposition products in KBr.
mechanism of Vaughan, et al.," accounts fairly well for the new bands that appear in the spectra of DTBP dispersions in KBr (Fig. 1). BANDSAPPEARINGIN
THE
DTBP
I
I
I
1
I
THERMAL DECOMPOSITION OF IN KBR
Band
Assignment
1705 vs 1300 w 1220 s 1170 w 910 m 785 w
C=O (mainly acetone) Methane Acetone Methyl ethyl ketone ( ? ) &Butyl alcohol something else Isobutylene oxide ( ? )
+
The gradual disappearance of the 0-0 stretch a t 875 cm.-' (Fig. 1 and Table I) follows first-
'"1 A
Pure Liquld IIbld, 12, 3 3 7 (195011 149'C
l27OC
I
I
I
2500
I 2 4 00
_I
IO4/ TOK.
Fig. 3.-First-order rate constant for scissioii of the peroxide link in di-t-butyl peroxide.
For the gas phase decomposition, Raley, Rust and Vaughan" found that k = 3 2 X 10lR ,-sg,looRT sec,-l
I t appears that dispersion in KBr disks may be a convenient and safe technique for following the decomposition of unstable solids and liquids. We have followed the decomposition of di-t-butyl perT I M E ( H R S ).
Fig. 2.-Kirieticd
a t a on scission of the 0-0 bond of di-t-butyl peroxide in KBr pellets.
FIRST-ORDER RATECONSTANT FOR SCISSION O F T H E 0-0 BONDOF DTBP IN KBR PELLETS Temp. ("'2.)
ins
k X 106 (set.-')
0.038
order kinetics for quite some while a t 127' (Fig. a), 127 2 96 and lower temperatures (at 149' the apparent 149 41 0 first-order constant steadily decreased with time). Plotted logarithmically against l/T°K. (Fig. 3), oxide in this way-in ordinary ovens without steeI these data give an activation energy of 38.0 kcal./ and concrete barricades-up to temperatures 40" mole above those of previous studies on the pure liquid. k = 1.45 X 101R e--IR,OO@lRT ~ ~ c , - l Only a few milligrams are needed in an average
April 20, 1957
THERMAL DECOMPOSITION OF DI-~-BUTYL PEROXIDE AND NITRATEESTERS
1795
100
.*$ v1
22
50
8
s? 0
3500
3000
2500 2000 1900 1800 1700 1600 1500 1400 1300 1200 11001000 900 800 Frequency (cm. -l). Fig. 4.-Spectra of ethyl nitrate and decomposition products in KBr.
700
100
.*i
w 3
3
50
& s? 0
4000
2000
3000
Fig. 5.-Formation
1800 1600 1400 1200 1000 Frequency (cm.-l). and decomposition of NOS- in potassium halides
disk (1.3 cm. X 0.25 cm.); a fairly complete kinetic study probably could be made with a gram, or less. The technique understandably works best when the kinetics are first order. TABLE I KINETIC DATAON THE THERMAL DECOMPOSITION OF D-tBUTYL PEROXIDE IN POTASSIUM BROMIDEPELLETS. ARSORBANCES OF THE 875 CM.-’ BAND
800
600
ture (Fig. 5). Accompanying the 1385 cm.-’ band, but weaker, is a sharp spike a t 835 cm.-’, which we have found with all ionic nitrates. These facts suggest that primary homolytic cleavage of the RO-NO2 bond (reaction 1) in KX (X = C1, Br or I) is followed by the reactionLa 2N02
+ K X = KNOa 4- [NOX]
(3)
We have verified spectroscopically that NOX is formed a t room temperature for X = C1 and Br. - log A hr. At higher temperatures (decomposition tempera0.228 23.13 6 109 tures ranged from 120-190°), it would presumably 0.784 119.38 6 109 exist largely as XZand nitric oxidela-although we 1.107 165.88 6 109 have not seen the latter, owing perhaps to its small infrared extinction coefficient, or more likely, 0.00 0.009 7 127 c 0.442 to its instability with respect to oxygen atom trans9.50 127 fer, to form COI (2350 cm.-l) and CO (2170 cm.-l) 1.030 21.50 7 127 (Fig. 4). 0.155 0.00 8 127 Heterolytic cleavage of the R-ONOZ bond to 0.591 10.0 8 127 form NOS- directly might give R-X as a by22.0 1.207 8 127 product. A search in the KBr region (400-800 0.283 0.00 min. 10 149 cm.-l) for the characteristic C-CI and C-Br .603 30.00 min. 10 149 stretching frequencies gave no evidence of this. ,806 60.00 min. 10 149 Further evidence for homolytic RO-NO2 cleavThermal Decomposition of Nitrate Esters in age, followed by reaction of NOz with the halide Potassium Halide Pellets.-Most striking in the matrix (according to (3)), comes from the twin obdecomposition of all nitrate esters in KBr was the servations (i) that the observed first-order rate appearance of a strong band a t 1385 em.-’ (Fig. 4). constant for the falling off of the nitrate ester The same band appears also in KCI and K I ; but bands of nitrocellulose (symmetrical stretch a t not in the spectra of heat treated films and smears 1280 em.-’, antisymmetrical stretch a t 1650 cm.-l) of nitrocellulose. It corresponds closely to the decreases markedly with time, suggesting some sort principal absorption band of ionic nitrates (Fig. 5), of interaction with the surrounding matrix; and and can be obtained quite easily by exposing pel(13) D. M. Yost and H. Russell, “Systematic Inorganic Chemistry”, lets of KCI, KBr or K I to NO3 a t room tempera- Prentice-Hall, Inc., New York, N. Y., 1946, p. 42. DTBPKBr pellet no.
Temp., OC.
Time,
HENRYA. BENTAND BRYCECRAWFORD, JR. I OXYGEN DONOR (OCCUPIEO
1-01. 79
aqueous nitric acid with respect to decomposition to nitric oxide and oxygen ? [ ~ X O L ( : I~ ( I 2NC) --I :$/2 cJ2 -4- 11x0
1
OXYGEN ACCEPTOR VACANT
- 7 ~
N20
N2
I
ZKSC&(c)
=
1F0(2!B) = +-3i.(i kc:il. 3 S O 4-2 1 2 0 2 K20 1F0(298j = +l53.1 k c d .
(-L)
+
The AF" difference, 133.1 - 36.6 = 116.5 kcal., comes largely from the fact that water is a much better sink for oxygen than K 2 0
+ Hz0 = K20 + 2H+(aq)
2E;'i:iq)
A F " ( 2 9 8 ) = +115.4 kcal.
(5)
(This contrasts sharply with the relatively small unitary free energy14-1b of solution of most ionic compounds, even insoluble cines such as barium sulfate (+16.6 kcal.)). Anhydrous potassium chlorate, on the other hand, can lose oxygen relatively easily KClO;j = KCI
+ 3/2
0 2
AFO(288) = -2S.3 kcal.
(6)
showing (cf. 4) that C1- and Nos- are quite stable in KC1 with respect to the alternative arrangement, 2NO c103-- O=(as K20).
+
KC103
+
+ 2 5 0 + K20 = 2KN03 + KCl AF"(298)
--I
=
-181.4 kcal.
(7)
I n assessing the relative stability of atomic conK X system, figurations accessible to the NO3X a halogen, we have found it quite useful to characterize a rearrangement such as (7) as the transfer of oxygen atoms. IVe say that KC103 (going to KC1) is a better oxygen-donor than KN03 0 100 000 :100 2-00 300 (going to NO K20); that NO KZO (going to 7 ( t " = 7' 0;) KNOB) is a better oxygen-acceptor than KC1 Fig ti ---Oxygen levels A F ' ( T ) = A F " ( 0 9 8 ) - ~ l S " ( 2 9 8 ) (going to KC103); or that oxygen atoms sponta= AF' ( t o = 7 + 2 5 ) for the reaction: oxygen-acceptor + neously leave the [KC103-KCl] level (written KC103 '/z 0 2 = oxygeti-donor. ( A F ' ( 7 ) = A F 0 ( 7 2 5 ) when when occupied and KC1 when vacant) for the [KN03- NO KzO] level (written KXOJ when AC,," = n ) occupied and NO K20 when vacant). We get in this way the left side of Fig. 6.16 (The corre(ii) that the rate of appearance of the 138.3 c m - ' sponding levels in aqueous solution are displaced upKO3-- band is greatest in K I matrices and least in 115.4 kcal. per H20 conKC1, this being precisely the order of the rates of ward approximately reaction of these halides with free NOZ. (This is also sumed per '/z 02.) $he best oxygen-donors appear by convention, evidence against participation of atmospheric oxyli e the best reducing agents in a table of oxidation gen in the pellet production of Nos-, confirmed by potentials, on top a t the left; the best oxygen-acIieating several pellets iiz vacuo.) ceptors at the bottom on the right (Zn, CO and HP It has been observed, also, that Nos- is itself an are included for comparison), Oxygen atoms tend unstable intermediate, albeit a relatively long lived to fall from upper-left to lower-right; in our example one. m'ith nitrocellulose, which has no strong from level 3 to level 12. The square root of the bands interfering a t 1385 cm.-', the yield of inor- oxygen partial pressure above level 3 is eiF3'/*27', ganic nitrate (defined as the absorbance of the that above level 1' 3 P ~ o - ~ : 3 e ~ ~ 1 2these ' / ~ ~ , two 1385 cm.-' band divided by the change in absorb- escaping tendencies being equal a t equilibrium. ance of the 1280 cm.-' band of nitrocellulose) Reaction 3 (for X = Br, and NOBr dissociated to decreased with increasing time in every observed NO Br2) corresponds to the tumbling of oxygen case. Yields, in addition, are a function of the atoms from level 6 to 7. temperature and the matrix, being greater in KBr Figure 6 has been extended off to the right in the than in KCl a t 122, 136, 156 and 186", but least of following manner. A F o ( T ) = A H ' ( T ) - TAall in K I a t 186' (after one minute)-although the So(T)zz AH'(298) - TAS'(298) = AF'(298) production of NO3- (Abs. a t 1385 +- initial Abs. ( T - 29S)AS0(298) = AF'(29S) - ~ A S ' ( 2 9 8 ) , a t 12SO) was greatest in KI, as observed previously T = t o - 25. SVhat has been plotted against r a t room temperature. in Fig. 6 is this quantity AF'(298) - ~ A S ' ( 2 9 8 ) , Thermal Decomposition of NO3- in Potassium abbreviated AF'(7). AF'(7) is precisely equal to Halide Pellets.-In the absence of a good sink (14) li. W. Gurncy. " I o n i c I'roccsscs in Solulii~n."lrcGraw--€Iill for left-olrer oxygen atoms, oxidizing agents such as I3ook Co., N e w York, S Y , 1 ! i T , R , Chap. 5 . KOa- are relatively stable. Anhydrous potassiuni D e n t , J . Phyr C i i e n ? . , 60, 123 (l!i5('Nj f l G j Cf ref 14, Glial, 7 , Sccliirli G!l nitr:ttcb, for c ~ u : i i i i p k , is itiuch more stable than
+
+
+
+
+ +
+
+
EXCHANGE BEHAVIOROF MONOSUBSTITUTED PEROXIDES
April 20, 1957
AFO (25' above r ) when ACPo for the reaction oxygen-acceptor l/Z 0 2 = oxygen-donor is zero. When ACpo# 0, AH' and A S o neverthelessincrease, or decrease, together, their increments tending to balance each other off in the second step above, which remains often a useful approximation. From Fig. 6 one sees that whereas lead nitrate decomposes around 300400° (level 10 to 4, forming PbO NO2 OZ), there do not appear to be acceptor levels below the highest KN03 donor level in KC1 doped only with KN03, a t least a t temperatures below the melting point of KCI. The instability of NOS- produced during the thermal decomposition of nitrocellulose probably arises from the presence in these instances of the low-lying hydrocarbon level 11. At 150" it is as good an acceptor as Hz. It should be mentioned, however, that a t relatively high temperatures (200-365') a KI pellet doped with a little KNOI produced a number of new bands after several days (Fig. 5). From the growth and decay of these bands with time a t several temperatures, it appears that the 2400 and 1750 bands probably belong t o the same molecule, or ion, 1685 and 950 to another, and 2150 and 1250 to still others. Spectrum of COz a t High Pressures.-Carbon dioxide appears to be formed in situ during the thermal decomposition of nitrate esters in KBr (Fig. 4). The dashed curve in Fig. 7 is a close-up of the 2350 cm.-l band produced in KBr in 55 minutes by the thermal decomposition of nitrocellulose a t 157". To reduce scattering, the pellet had been recompressed to 12,500 p.s.i. For comparison, the spectrum of atmospheric COZ is shown, drawn to the same total integrated intensity (Slog (Io/I)). Recompression to 125,000 p.s.i. (solid curve) continued to diminish the relative intensity of the high frequency branch ; the integrated intensity, however, was not noticeably affected. Cooling to liquid nitrogen temperatures further sharpened the band, again with little change in the total inte-
r t
+
+
I
I
I
I
I
I
I
I
I
1 1
-
60
56-
52 -
-
+
[CONTRIBUTION FROM T H E
I
1797
48-
40 36 -
44
I
d
\ 0
-
32-
t!-
2 28-
0
-
24
-
20
-
16-
12-
04 08
FREQUENCY (cm.-I),
Fig. 7.-Spectra
of COz a t high pressures.
grated intensity. These effects are reversiblewarming tends to restore the normal band contour-, reproducible, in KBr and KC1, and relatively permanent a t room temperature. Acknowledgments.-The authors wish to thank Miss Ann McHale for her helpful assistance in many phases of this work, and Drs. John Overend and William Fateley for their many helpful suggestions. MINNEAPOLIS 14, MINN.
METCALF CHEMICAL LABORATORIES, BROWNUNIVERSITY]
The Exchange Behavior of Monosubstituted Peroxides BY GERRITLEVEY,'DONALD R. CAMP BELL,^ JOHN 0. EDWARDS AND
JANE
MACLACHLAN
RECEIVED DECEMBER 7 , 1956 A study of possible isotope exchange between four monosubstituted peroxides and the corresponding anions has been carried out. I n three cases (sulfate, benzoate and acetate), no exchange was observed even a t the temperature of rapid peroxide decomposition. I n the phosphate case, no way of separation of peroxide from anion was found. Mechanistic conclusions are pointed out.
The exchange behaviors of hydrogen peroxide and peroxydisulfate ion have been studied extensively. Hydrogen peroxide does not appear to exchange with either labeled H20 or OH- under a variety of condition^.^ The exchange of radio(1) Visiting Summer Research Associate, 1956. (2) National Science Foundation Predoctoral Fellow, 1956-1957. (3) (a) R . S. Winter and H. V. A. Briscoe, THISJOURNAL, 73, 496 (1951); (b) P. Baertschi, Erperientia, 7 , 215 (1951); (c) J. Halperin 74, 380 (1952); (d) M. Dole, G. Mucand H. Taube, THISJOURNAL, how, De F . P. R u d d and C. Comte, J . Chem. P h y s . , 20, QG1 (1952).
sulfate ion with peroxydisulfate ion also proceeds very slowly if a t all.4 I n view of the more reactive nature of the unsymmetrical peroxides such as Caro's acid, it was felt worthwhile to find out whether exchange could be detected with them. Results with Caro's acid (4) ( a ) I. M. Kolthoff and I. K . Miller, THISJOURNAL, 7 8 , 3055 (1951); (h) R. L. Eager and K . J. McCallum, Con. J. Chem., 32, 692 (1954); (c) P.C.Riesebos and A. H. W. Aten, Jr.. THISJOURNAL, 74, 2440 (1952); (d) H . Elkeles and C. Brosset, Svensk. K e m . T i d s k r . . 66, 26 (1953).
