354
J. Phys. Chem. 1981, 85,354-360
depolarization ratios of 0.65 f 0.1 were found for each of the three strong anion bands at 1586,1475, and 957 cm-l. The origin of the comparable depolarization ratios for the three modes is tentatively identified as the rapid switching of B; between two D2h structures of different form but nearly equal energies. It is presumed that, consistent with calculational results showing rather minor energy barriers between the Dul structures,l* the structure fluctuates at a rate approaching the vibrational time scale such that an assignment of the vibrational modes to simple symmetry classes is impossible. This tentative explanation will be pursued in subsequent studies in the light of theoretical analyses that are applicable to this case.16 K'B; Ion Pair. The steady-state relationship proposed for the laser-irradiated sample requires the presence of the K'B; ion pair as well as the isolated anion radical. Further, it is clear that the ion pair also has absorption bands near the laser excitation frequencies, though the assignment of the 650-nm absorption is not without some doubt. Thus, a resonant scattering from the ion pair is anticipated, and the weak subset of Raman bands, which is enhanced by increasing the excitation wavelength, is tentatively assigned to the K+B; ion pair. There is a definite argument for such an assignment independent of this preresonance behavior. The spectrum more closely resembles that of neutral benzene than does the stronger Raman subset that has been assigned to the isolated B;. This is anticipated for the ion pair, particularly if the symmetry is not disrupted by the cation, Le., the cation is positioned on the molecular axisa2 Back-donation of charge to the cation, which is apparently complete for sodium, is expected to be significant for potassium and rubidium as well and thus should reduce the static JahnTeller distortion of the ring, along with the frequency shifts resulting from bond strength reduction from the radical electron. The infrared data for the unbleached films is attributed to M'Bz- and is therefore included in Table I1 where the (16) M. Pawlikowski and M. 2. Zgierski, Chem. Phys. Lett., 48 201 (1977).
vibrational assignments for the ion pair are indicated, based on a c6"molecular symmetry.
Conclusions The resonant Raman spectrum for the benzene anion isolated in a benzene matrix can be interpreted by assuming that a major static Jahn-Teller distortion exists. The magnitude of this distortion, required to produce the observed splitting of the eZgvibrational modes, is similar to that predicted by Gaussian orbital-based quantumchemical calculations.'" Both the molecular orbital calculations and the vibrational data are consistent with ring bond-length differences of -0.08 A for the distorted structures. This means that one of the two probable DZh structures of similar energy has a ring structure that is nearly equivalent to that of the ring portion of the TCNQ monoanion. Though the Raman spectrum shows discrete bands, they are moderately broad (15-30 cm-l) and are tentatively interpreted as reflecting the rapid interconversion between D2h quinonoid and elongated structures on a time scale approaching that of the molecular excitation. In particular, some such viewpoint must be invoked to understand the comparable depolarization ratios for the three Raman active ring stretching modes. A second reaction product of benzene with metal is evident from the Raman spectra. Since it is reasonable that a steady state develops in the samples exposed to the laser beam, with the MfB; species participating as one component, the Raman bands that indicate the presence of a second product have been assigned to the ion-paired benzene anion. If this assignment is correct, then it follows that the 650-nm absorption band, reported by Gardiner and observed in our deposits as well, is produced by M+B; since the resonant character of the Raman scattering from this second species increases with increasing laser excitation wavelength. The assignment of such a low-energy, ion-pair electronic excitation is not firm but the chargetransfer excitation to the "no-bond" state is a likely candidate. Acknowledgment. The research was supported by NSF grants CHE 77-09653 and CHE 79-25567.
Electron Spin Resonance Study of the Reaction of Aromatic Hydrocarbons with Oxygen I. C. Lewis" and L. S. Slnger Union Carbide Corporation, Carbon Products Division, Parma Technical Center, Cleveland, Ohio 44 10 1 (Received: August 28, 1980)
Electron spin resonance was used to study the free-radicalintermediates formed during the reaction of polynuclear aromatic hydrocarbons with oxygen. The treatment of a variety of hydrocarbons with O2 at 275-320 "C in an inert solvent gave rise to colored solutions which generally exhibited intense ESR signals. Proton hyperfine spectra could be resolved for approximately half of the radicals. The high g values of between 2.0029 and 2.0040 confirmed the presence of chemically bound oxygen in the radical structure. The g values show a linear relation when plotted against either the energy of the lowest vacant molecular orbital or the polarographic half-wave reduction potential of the parent hydrocarbon. Preliminary analysis of some of the hyperfine spectra indicates that the radicals are stable aryloxy radicals produced by substitution of hydrogen by oxygen at a reactive site in the molecule.
Introduction Polynuclear aromatic hydrocarbons are known to react directly with oxygen to form a variety of oxidation products. These reactions are usually catalyzed photochemi0022-365418112085-0354$01 .OO/O
cally,l by the addition of base2 or by strong acidsa3 Free radicals have often been postulated as intermediates in (1) C. DuFraisse, Bull SOC.Chim. Fr., 6,422 (1939).
0 1981 American Chemical Society
Reaction of Aromatic Hydrocarbons with Oxygen
these oxidation reaction^.^ Air oxidation has also been shown to effect important chemical changes in complex polynuclear aromatic hydrocarbon mixtures, such as coaland petroleum-derived tars and p i t ~ h e s . ~ There have been several studies utilizing electron spin resonance (ESR) to characterize the free-radical products from the reaction of aromatic hydrocarbons with oxygen. Yoshida et a1.6 heat treated anthracene in the presence of air from 200 to 300 "C in the cavity of an ESR spectrometer and observed a partially resolved proton hyperfine spectrum which they attributed to an anthryloxy radical. Forbes and Robinson' observed a well-resolved ESR hyperfine spectrum for the hydrocarbon azulene when heated in air at 137 "C. This spectrum was also attributed to an aryloxy radical. The hydrocarbon 3,4-benzopyrene also gave an ESR spectrum when heated in oxygen which differed from that obtained without This paper presents an ESR study of the thermally induced reaction of oxygen with a variety of polycyclic aromatic hydrocarbons. The results provide a further application of the ESR technique for characterizing stable radical intermediates in organic reactions. Experimental Procedure The polynuclear aromatic hydrocarbons were obtained from commercial sources and used without additional purification. The hydrocarbon o-terphenyl (Aldrich Chemical Co.) was employed as an inert solvent. This compound exhibits a melting point of 57 "C and a boiling point of 332 "C at 760 mmHg. The polynuclear hydrocarbons were mixed with the o-terphenyl in ratios of from 15 to 1:20. The mixtures were placed in standard Pyrex 4-mm i.d. NMR sample tubes and heated in a Varian standard heater cavity while oxygen was bubbled through the solution. The standard heat-treatment temperature was 275 "C except for azulene, which was heated at 200 "C, and 3,4-benzopyrene (260 "C). The oxygen addition was continued until an ESR signal was observed, generally -20 min. After appearance of an intense signal, the sample tubes were removed, evacuated by the usual freeze-thaw technique to remove the residual oxygen, and sealed under vacuum. The ESR spectra were then measured by utilizing a conventional 10-GHz ESR spectrometer employing superheterodyne detection at a microwave power of 0.1 mW. The temperature was varied until the maximum resolution was obtained prior to recording of the spectra. Technique of g-Value Measurement Accurate g values were determined by measuring in quick succession proton and microwave frequencies employing an EIP, Inc., frequency counter Model 360 A, equipped with Autohet converter Model 362 A. The Varian Associates' Model F-8A fluxmeter was used to obtain the proton signal from a water sample in the probe mounted directly behind the TE-102 microwave cavity. A comparative method was used in which the g value for 0.001 molar solution of perylene in sulfuric acid at room (2) H. Pines and W. M. Stalick, "Base Catalyzed Reactions of Hydrocarbons and Related Compounds", Academic Press, New York, 1977, p 508. (3) A. Zinke and E. Unterkreuter, Monatsh. Chem., 40, 405 (1919). (4) J. 0. Hawthorne, K. A. Schowalter, A. W. Simon, and M. H. Wilt, Adu. Chem. Ser., No. 75, 203 (1968). (5) J. B. Barr and I. C. Lewis, Carbon, 16, 439 (1978). (6) T. Yoshida, Y. Ueno, and S. Wakabayashi, Kogyo Kagaku Zasshi, 72, 338 (1969). (7) W. F. Forbes and J. C. Robinson, Nature (London),214,80 (1967). (8) W. R. Forbes and J. C. Robinson, Nature (London), 217, 550 (1968).
The Journal of Physical Chemistty, Vol. 85, NO. 4, 198 1 355
GAUSS
Flgure 1. ESR spectra for free radicals from the reaction of (a) anthracene and (b) pentacene.
i
O2with
1.0 GAUSS
,
Figure 2. ESR hyperfine spectra for free radicals from the reaction of O2with (a) pyrene and (b) acenaphthylene.
temperature was assumed to be 2.002 569. This is the value reported by Segal et al.? corrected for the Allendoerfer correction.1° As shown by Segal et al.,9 the g value (g,) of an unknown can be obtained, to good approximation, by the expression g x = gstd(vmic/lp)x/(Vmic/Vp)std
where vfic/v is the ratio of microwave to proton frequency, and gsais t i e g value of the perylene standard. After optimum placement of the microwave cavity had been determined by a careful field mapping, and attention was given to reproducible field cycling procedures, g values could be reproduced to within 1 or 2 ppm. Experimental Results ESR spectra were observed for 20 polynuclear aromatic hydrocarbons after heat treatment in the presence of oxygen. The compounds and the ESR results are listed in Table I. Many of the ESR spectra contained resolved proton hyperfine lines. The number of hyperfine lines observed and the experimental line widths are also given in the table. The measured g values are listed in the last column of the table. Several of the compounds listed at the end of the table gave signals which were either too weak or asymmetric to permit g-value measurement. No ESR spectra was obtained for the hydrocarbons fluorene, triphenylene, and tetrabenzonaphthalene after 30 min at 275 "C. The hydrocarbon 5,6,11,12-tetraphenylnaphthalene (rubrene) gave an ESR signal only after heating for several hours with O2 at 320 "C. (9) B. G. Segal, M. Kaplan, and G. K. Fraenkel, J. Chem. Phys., 43, 4191 (1965). (10) R. 0. Allendoerfer, J. Chem. Phys., 55, 3615 (1971).
356
The Journal of Physical Chemistry, Vol. 85, No. 4, 1987
Lewis and Singer 200
. 7
%/
7-
150 Lo
0
x 01
w
3
J
Figure 3. Second derivative ESR hyperfine spectrum (half spectrum) for free radical from the reaction of perylene with 02.
100
$ 01
.-C t-
k
5,
50
0
G&+
i
o*
2.0 GAUSS
,
\'
n
01 0 2 0 3 0 4 0 5 06 07 MOLECULAR ORBITAL ENERGY COEFFICIENT, - m m "
Figure 8. Correlation of g shifts for radicals prepared by reaction of aromatic hydrocarbons with oxygen, and HMO coefficient of lowest vacant orbital of parent hydrocarbon. For identification of compounds, see Table I.
A
Figure 4. ESR hyperfine spectra (haif spectra) for free radicals from the reaction of O2with (a) 9,lOdihydroanthracene and (b) 9,9'-bianthryl.
&+
00
o2
Flgure 5. ESR hyperfine spectrum (half spectrum) for free radical from the reaction of 3,4-benzopyrene with 02.
The appearance and overall resolution of the ESR spectra varied for each compound. Figure 1 presents the ESR spectra measured for oxidation products of anthracene and pentacene. Both spectra consist of a series of equally spaced hyperfine lines superimposed on a broad background. The ESR spectra for the pyrene- and acenaphthylene-derived radicals are reproduced in Figure 2. The pyrene spectrum contains two separate groups of hyperfine lines indicative of the presence of of a single large proton hyperfine splitting. Well-resolved but complex ESR spectra were obtained for the radicals from perylene, 9,10-dihydroanthracene, 9,9'-bianthryl, and 3,4-benzopyrene. These spectra are reproduced in Figures 3-5, respectively.
Discussion of Results It is clear that stable free radicals are produced when polynuclear aromatic hydrocarbons are heated in the presence of oxygen. The key question is the identity of these radicals intermediates. The measured g values all range between 2.0029 and 2.0040, demonstrating the presence of oxygen in the free radical. The g values also vary with aromatic structure. Segal et al.9 have shown the general correlation of g values for aromatic hydrocarbon radical anions and cations with the energy level coefficient of the Huckel molecular orbital
occupied by the unpaired electron. This correlation is in agreement with the theory of Stone.ll In Figure 6, the g values for the oxygen-containing radicals measured in this study are plotted vs. the energy coefficient of the lowest vacant molecular orbital for the parent hydrocarbon.12 The plot is expressed in terms of Ag, or the shift in g value from the free electron g value (2.002319) vs. the energy level coefficient -mm+'. A good correlation is found for all of the radicals with the exception of coronene for which the vacant orbital is degenerate. It is significant that the agreement is much poorer when the g-value shifts are plotted vs. the coefficient of the highest filled orbital, a result which can be tested for the nonalternant hydrocarbons which are included. In Figure 7, the g values are plotted against the polarographic reduction potentials of the parent hydrocarbon.12 The g-value results and correlations strongly suggest that the radicals retain the original aromatic structure but include a substituent oxygen atom. A definitive identity of the radicals could be provided by complete analysis of the proton hyperfine spectra and accurate determination of the coupling constants. Many of the spectra are too poorly resolved to permit a direct analysis. Several of the better-resolved spectra are complex probably because of the unsymmetrical nature of the radical structure. Analysis of these spectra is being tried by using computer-assisted schemes. We have, however, been able to obtain proton coupling constants for some of the radicals, and the results are consistent with an aryloxy structure (I) in which oxygen b I
Fi 1
(11) A. J. S:one, Mol. Phys., 6, 509 (1963). (12) A. Streitweiser, Jr., Ed., "Molecular Orbital Theory for Organic Chemists", Wiley, New York 1961.
The Journal of Physical Chemistry, Vol. 85, No. 4, 198 1 357
Reaction of Aromatic Hydrocarbons with Oxygen
6
200
:*: I50 6' 7'@&3
YI
e
5'
x
0
9'
4'
I1
4
w
maining splittings likely arise from the 1, 3,6,and 8 protons. Anthracene. The radical derived from the reaction of anthracene and oxygen is likely an oxyanthracene radical (111) similar to that proposed by Yoshida et al? The poor
3 :3 100 0,
c ._
t-
LL
. I
m
50
0 0
5
-€
'92
1.0 1.5 2 .o versus Hg (2-METHOXYETHANOL)
Flgure 7. Correlation of g shifts for radicals prepared by reaction of aromatic hydrocarbons with oxygen, and polaro raphic half-wave reof parent hydrocarbon.'' For identification of duction potential (-€'I2) compounds, see Table I.
(a' I L \
\ V I
I V
I
III resolution of the spectrum (Figure la) prevents the unambiguous assignment of any coupling constants. Pentacene. The ESR spectrum in Figure l b is not sufficiently resolved for analysis, but a reasonable assignment for the radical would be the oxypentacene structure (IV) in which oxygen is substituted at the most reactive 6-position.
QIGJfpQ '
'
2 GAUSS
'
Flgurr 8. Comparison of experimental and computer-simulated second-derivative spectra (half spectrum) for radical from reaction of 9,9'-bianthryl with oxygen: (a) experimental secondderivative spectrum; (b) computed spectrum: 1 H, a , = 0.20 G 2 H, a 2 = 3.40 G; 2 H, a 3 = 3.06 0 ; 2 H, a 4 = 1.03 G ; and 2 H, a 5 = 0.80 G .
has substituted for a hydrogen atom at a reactive ring site. The unpaired electron can be readily delocalized into the aromatic ring system through resonance interaction. Our analysis of the spectra is consistent with the aryloxy structure I. These results are discussed for the various radicals individually. 9,gl-Bianthryl. The hyperfine spectrum of the radical from the reaction of 9,gl-bianthryl with oxygen has been reduced to the following coupling constant^:^^ 1 H, al = 0.20 G; 2 H, = 3.40 G ; 2 H, a3 = 3.06 G ; 2 H,a4 = 1.03 G; 2 H,a5 = 0.80 G. A comparison of experimental and simulated spectra is shown in Figure 8. This assignment is consistent with structure I1 for the radical. This radical is formed by the substitution of oxygen for a hydrogen at the reactive 9-position. The single small splitting of 0.20 G can resonably be assigned to the proton at the 9'-position. The large splittings, 3.40 and 3.06 G, are most likely attributable to the 2, 4, 5, and 7 protons while the re(13) L. S. Singer, I. C. Lewis, T. Richerzhagen, and G. Vincow, J.Am. Chem. SOC.,75, 290 (1971).
0
IIL 3,4-Benzopyrene. The resolved proton hyperfine spectrum measured for 3,4-benzopyrene when heat treated in the absence of the air was attributed to an azulene-type hydrocarbon radical? We also obtained a spectrum when 3,4-benzopyrene was heated at 200 "C under vacuum. The measured g value of 2.002 64 confirms the hydrocarbon nature of the radical. Our spectrum for the reaction of 3,4-benzopyrene with oxygen differs appreciably from the radical measured in the absence of oxygen. The spectrum in Figure 5 can most likely be assigned to an oxybenzopyrene radical (V). The oxybenzopyrene radical V has
0a
P been shown to be extremely stable and forms readily by dissociation of hydrogen from 6-hydroxybenz~pyrene.'~ The resolution for the published ESR spectrum obtained for V was not sufficient to permit an unambiguous comparison with the spectrum in Figure 5. (14)K. P. Mishra, Y. Nosaka, K. Akasaka, C. Nagata, and H. Hatano, Biochim. Biophys. Acta, 520, 679 (1978).
358
Lewis and Singer
The Journal of Physical Chemistty, Vol. 85, No. 4, 7987
TABLE I: Survey of ESR Results Obtained for the Reaction of Aromatic Hydrocarbons with Oxygen concn in meas! no. of hyperfine aromatic hydrocarbon o-terphenyl temp, C lines LW, G g value 1. 9,lO-dihydroanthracene
1:lO
105
66
0.1
2.003 39 (tO.00003)
2. anthracene
1:5
85
20
1.0
2.003 70 ( ~ 0 . 0 0 001)
3. tetracene
1:20
120
NR
4.0
2.003 31 (rO.OOO 10)
4. pentacene
1:lO
120
31
0.3
2.002 95 (tO.OOO 01)
5. 9,9’-bianthryl
1:60
120
100
0.12
2.003 38 (tO.00002)
6. perylene
1:20
130
60
0.09
2.00322
7. pyrene
1:lO
145
46
0.08
2.003 99 (.+O.OOO 01)
1:lO
265
0.3
2.00335 (+O.OOO 10)
1:lO
80
2
2
2.003 8 (tO.0004)
10. 3,4-benzpyrene
1:20
75
166
0.08
2.003 48 (tO.0000 2 )
11. 1,2-benzpyrene
1:lO
95
30
0.1
2.004 02
12. azulene
1:20
100
NR
4
2.003 13 (*O.OOO 05)
* @
8. 9,lO-dimethylanthracene
25 + broad background
CH3
9. fluoranthene
a
The Journal of Physical Chemistry, Vol. 85, No. 4, 198 1 350
Reaction of Aromatic Hydrocarbons with Oxygen
TABLE I (Continued) aromatic hydrocarbon
concn in measd o-terphenyl temp, "C
no. of hyperfine lines LW,G
g value
4
2.003 8 5 ( ~ 0 . 0 0 004)
0.4
2.003 32 (iO.00002)
NR
3
2.003 78 (+O.OOO 0 5 )
120
NR
3
2.003 84 ( ~ 0 . 0 0 005)
1:20
180
NR
5
2.003 4 (*O.OOO 1)
1:lO
265
19
0.3
2.003 2 (*O.OOO 1)
1:lO
265
19
0.4
2.003 26 (+O.OOO 06)
20. 1,12-benzoperylene
1:lO
155-200
- 30
0.1
21. 9,lO-methylenephenanthrene
1:10
13. 1,2-benzanthracene
1:20
80
14. anthanthrene
1:lO
100
15. 1,2:5,6-dibenzanthracene
1:lO
120
16. 1,2:7,8-dibenzanthracene
1:lO
17. coronene
&@
NR
-resolution 30, partial
@ 18. acenaphthylene
@ 19. biacenaphthylidene
% H
GO
80
Pyrene. It is evident from the spectrum in Figure 2a that the radical contains a single proton with a large hyperfine splitting. The radical (VI) produced by substitu0
I\zf
tion of oxygen at the most reactive 1-position would be
NR
5
expected to exhibit a large proton splitting from the adjacent hydrogen. Perylene. The well-resolved but complex hyperfine spectrum (Figure 3) for the perylene derived radical can likely be assigned to either aryloxyradical structure VI1 or VIII. These positions are the most reactive in the perylene molecule. Acenuphthylene. The ESR spectrum observed from the oxidation of acenaphthylene consisted of a series of hyperfine lines superimposed on a broad unresolved background signal. The resolved hyperfine spectrum gave the appearance of a quintet of triplets consistent with a group of four and two equivalent protons in the radical. The apparent coupling constant assignment is the following:
360
J. Phys. Chem. 1981, 85,360-367
radical has been shown to form during pyrolysis of biacenaphthylidene.16 9,10-Dihydroanthracene. The absence of a central peak in the ESR spectrum from the oxidation of dihydroanthracene (Figure 4a) indicates a hyperfine splitting by an odd number of protons in the radical. The most reasonable structure is the aryloxy dimer radical X.
m
Hm
4 H,C Z ~= 6.2 G; 2 H,a2 = 1.9 G. Since acenaphthylene is known to polymerize readily when heated to 200 O C , 1 6 a reasonable structural assignment for the radical would be the entity IX with an ace-
8 0 0 H H
x
Tx naphthylene oxy moiety incorporated in a polymer chain. The oxy radical structure contains six protons consistent with the ESR analysis. The acenaphthylene dimer, biacenaphthylidene, gave a similar spectrum but exhibited a number of additional lines believed to be due to the perinaphthyl radical. This (15) M.Kaufman and A. F. Williams, J. Appl. Chem., 1,489 (1961).
Conclusions Stable free radicals are produced when polynuclear aromatic hydrocarbons are heat treated in the presence of oxygen. The measured g value and the analysis of some of the proton hyperfine spectra indicate that these radicals are aryloxy radicals in which oxygen is substituted for hydrogen at a reactive ring site. The oxidation mechanism of polynuclear aromatic hydrocarbons is complex and is known to produce a variety of oxygenated products including hydroperoxides and quinones. The results of these ESR studies demonstrate that aryloxy radicals play an important role as intermediates in the overall oxidation process. (16) L. S. Singer and I. C. Lewis, Carbon, 2, 116 (1969).
Semiempirical Hot Atom Theory. 1. Initialization and Application S. Aronowitz,*l S. Chang, and T. Scattergoodl Extraterrestrial Research Division, Ames Research Center, NASA Moffeff Field, California 94035 (Received: October 23, 1980)
A novel approach to the theoretical assessment of the kinetics of systems containing both hot and nonhot reactions is proposed. The method is tested explicitly on experimental results of Martin and Willard for the (DBr + CHI) and (HBr + CD4)systems. The simulated data are in excellent accord with the experiments.
Introduction In the course of investigating the chemical dynamics of planetary atmospheres, it became evident that classes of reactions might be initiated by hot hydrogen atoms, that is, by hydrogen atoms whose kinetic energy exceeds the thermal kinetic energy. These hydrogen atoms, formed by photodissociation of simple molecules such as ammonia, methane, hydrogen suifide, and phosphine, have sufficient kinetic energy to form free radicals by hydrogen abstraction. Therefore, in systems containing these reactants it is necessary to examine this mechanism for radical production and then to follow the subsequent interaction of these radicals, as well as those formed directly by photo(1) National Academy of Science Research Associate.
dissociation,with each other and with other undissociated reactants. Our investigations were motivated by the desire to bridge the gap between laboratory conditions and planetary environments. However, it is usually too costly in effort and money to attempt accurate laboratory simulations of primitive atmospheres (e.g., that of Jupiter), because either too little is known about such atmospheres or the environments are impossible to duplicate in the laboratory. Our efforts then turned to modeling approaches. A probabilistic kinetic theory of hot-atom reactions was formulated some time ago by Wolfgang? The theory was used in a correlative rather than a predictive fashion be(2) Wolfgang, R. J. Chem. Phys. 1963,39,2983.
0022-365418 112085-0360$01.0010 0 1981 American Chemical Society
