Luminescence of Ionic Slates of Alkane Solids
2081
e Probe Analysis of Ionic States of Cyclopentane, Cyclohexane, and m-HexavrrR Solids under Low-Energy Electron Impact irnsthy Huang and William H. Hamill” De,wartnient of Chemistry and the Radiation Laboratory,’ University of Notre Dame, Notre Dame, Indiana 46556 (Received April 2, 1974) Publication costs assisted by the U. S. Atomic Energy Commission
Thin films (-200 A) of cyclopentane, cyclohexane, and n-hexane at 60°K doped with 1%benzene or toluene luminesce under low-energy electron impact. Bands appear at -200, -275, and -380 nm due to alkane Pnuoreiicence and to the aromatic hydrocarbon fluorescence and phosphorescence, respectively. For benzene in cyclohexane the 370-nm luminescence excitation spectrum has peaks at 7.8, -10, -13, 14.5, 15.9, 17.6, and 19.1 eV. The spectrum for toluene in cyclohexane is very similar and comparable spectra were obtained for n-hexane. For benzene in cyclopentane the excitation peaks appeared at 8.5, 9.3, 11.0, 11.9, 13.5, 14.7, and 16.7 eV. The principal mechanism of excitation appears to be ionization of a molecule of the matrix, posit,ive hole migration and trapping by the impurity, charge recombination to form an excited molecule of the additive, and luminescence. The luminescence excitation spectrum of the probe detects only those posit iare-hole states which can migrate and be trapped prior to recombination. The spectra indicates that the energy band width for migration is rather narrow.
Introduction Characteri3tic energy losses measured by slow electron impact on solid alkanes are usually poorly resolved, both for excited neutral states and for ionic state^.^,^ A comparable effect i s observed optically. The luminescence bands of alkane solids under electron impact which appear a t -200, -290, -370, and -490 nm provide some additional i n f ~ r m a t i o n .The ~ 200-nm band has been thoroughly investigated by Lipsky’s group5 and will not be considered further. The other ennissions of alkanes under electron impact arise from ionic states, and are probably due to fragments, which may include Their excitation spectra contain no resolved structure. A further possible approach to the problem of resolving the electronic structure of alkane solids is to add a positivehole trap which will generate a suitable recombination luminescence. ‘The exckation luminescence spectrum would then describe the delocalized positive ion (or hole) energy levels. Preliminary experiments with 1% benzene in cyclohexane showed adequate intensities of benzene fluorescence and phosphorescence and well-resolved luminescence excitation spectra. Fragmented ions would contribute little to benzene luminesceiice since they are localized. They do contribute an overlapjping luminescence, but it is weak for excitation below -18 cVV. Experimental Section The apparatus and general procedures have been des ~ r i b e d .Sainples ~ ~ ~ ? ~of 1mol % benzene or toluene in cyclopentane. cyclohexane, or n-hexane were admitted LO the high-vacuum system and maintained a t 2 X Torr for 480 sec. Film thickness would approximate 200 A. The substrate was held a t 601°K unless otherwise specified. The photon-counting period was 14 sec a t each energy. Firstderivative excitation spectra, dL/cl V us. eV, were obtained from the luminescence intensity L by the photon count difference at voltages V and V 0.5. A high-intensity Bausch and Lomb monochromator was
+
used for all measurements except those for benzene in cyclopentane for which a Corning 7-54 filter isolated the combined 280-nm fluorescence and 380-nm phosphorescence.
Results The luminescence spectra for n-hexane and cyclohexane doped with benzene a t 15-eV incident electron energy appear in Figure 1. The 200-nm band is due to alkane fluorescence. The bands at -275 and -380 nm correspond to the fluorescence and phosphorescence of benzene, although a weak admixture of alkane emission^^,^ cannot be excluded. The spectrum of toluene-doped cyclohexane for similar conditions appears in Figure 2. There is no evidence of excimer emissions and the activators are presumed to be homogeneously dispersed. This is to be expected for deposition at 60°K The excitation spectrum of 390-nm emission by electron impact for toluene-doped cyclohexane appears in Figure 3 (above) together with that for 370-nm emission from benzene-doped cyclohexane (below). The results from photoionization spectroscopy (Ps) of cyclohexane7 are shown by bars at the top of Figure 3, but with an arbitrary zero of energy. The electron energy scale is zeroed to the first (injection) peak of the energy loss s p e ~ t r u mThis . ~ choice will be correct if the electron affinity of the solid is zero and if there is negligible charge trapping in the film. The ionization potentials from photoelectron spectroscopy can be brought into register with the excitation peaks for electron impact by applying a constant shift of -2.0 eV. There is one “extra” excitation peak (or shoulder) for the solids a t 14.5 eV, but this is not unexpected for electron impact because of different selection rules. The excitation spectrum for 280-nm luminescence of benzene-doped cyclohexane in Figure 4 can be correlated with the results of photoelectron spectroscopy except that the luminescence associated with the ground state of C&12+ is even weaker than before and a peak cannot be located reliably. Correlations are based therefore. on the three strongest peaks a t 10.3, 15.9, and 17.0 eV. There is The Journaiof Physical Chemistry, Vol. 78, No. 21, 1974
Timothy Huang and William H. Hamill
2082
40
8
-4
-3 -2
-I
-0
S
206
L
300
2
400
d
Wovelength, nm
Figure 1. Emission spectra of 1% benzene in cyclohexane (0)and in nhexane (0)“ both a t 60°K and at 15-eV incident electron energy.
Electron Energy, eV
Figure 3. Excitation spectra of 390-nm emission from 1% toluene in cyclohexane (0)and 370-nm emission from 1 % benzene in cyclohexane (0).Data from photoionization spectroscopy are shown by bars after shifts of -2.0 eV.
Figure 2. Emission spectra of 1% toluene in cyclohexane at 60°K and at 15 eV incident electron energy.
“extra” structure at -14.6 eV, as in Figure 3. The results for cyclohexane are summarized in Table I. Considered together, these three excitation spectra support the interpretation that they correspond to the ionic states of cyclohexane. Their energy levels lie about 2.0 eV lower in the solid state than under vacuum. The excitation spectrum of benzene-doped cyclopentane appears in Figure 5 . Luminescence below -8-eV electron energy was also observed in n-hexane but not in cyclohexane. Possible correbations with data from photoelectron spect~oscopyappear as bars at the top. Tho excitation spectrum of 390-nm luminescence from benzene-doped n-hexane appears in Figure 6. The peak at 6.3 eV coincides with a resolved resonance in the characteristic energy loss spectrum.2 The other peaks occur at 8.8, 10.1, 11.8, 13.3, 15.3, and 16.8 eV. These results do not correlate with those for excitation of benzene phosphorescence The Journalof Physical Chemistry, Vol. 78, No. 21, 1974
1
I
I
8
IO
1
I
I
12 14 16 Electron Energy, eV
I
18
2
Figure 4. Excitation spectrum of 280-nm luminescence for 1 % benzene in cyclohexane.
in cyclohexane, This supports the interpretation that they represent states of the lattice. There are no results from photoelectron spectroscopy for comparison.
Discussion The luminescence spectra in Figures 1 and 2 do not permit an estimate of the extent of quenching 200-nm alkane fluorescence by benzene. In liquid cyclohexane 1%of benzene quenches about 80% of the fluorescence.8 Whatever
Luminescence of Ionic States of Alkane Solids
2083
TAB1,E I: Ionixrationt Potentials of Cyclopentane and Cyclohexane (eV) ________-_I
c-6 a"la (solid, c6&,
+
280nm 370 nm) 0.5 9.3 11. .o Ili .9 13.5 14.7 16.7
c-CsHlob (vapor, PS)
Shift, c-CaHlo
10 "49
2 .o
13 89 15.91
2.0 2.4
18.27
1.5
C-CsHi2~ (solid, C - C ~ H I ~ ~C-6eHiad C6H&H3, (solid, C6H6, (solid, C6H6, 390 mm) 370 nm) 280 nm) 7.8 -10.3 -12.8 14.7 16.3 17.8 -18.8 19.9 23.3 25.9
7.8 -10 -13 -14.5 15.9 17.6 19 .o
-9 10.3 -13.2 -14.6 15.9 17 .O 19 .o
c-C6H12* (vapor, PS)
(vapor, PS)
c-c 6H12 e
9.81 (11.88) (12.74) 14.50
9.88 (11.33) 12.22 14.37
(18.04) (19 .34)
19.43
Shift c-CsH1, 2 .o
1.9 -1.4 -42
.o
2 .o
From Figure 5. 6 ]?ram ref 7. Values in parentheses are uncertain. z From Figure 3. From Figure 4. e M. I. Al-Joboury and D. W. Turiier, J . Chern. Soc., 4434 (1964). 25
20
7 I5
0 x
>
-. v
J
.u IC
5
Electron Energy, eV
Figure 5. Excitation spectrum of the combined fluorescence and phosphorescence of benzene in cyclopentane.
the effect may be in the solid state, it need not concern us further. Alkane emissions a t -280, -350, and -490 nm are excited strongly only at -18 eV or more.2 Consequently the -275- and -~380-nrn bands of Figures 1 and 2 under 15-eV excitation are due entirely, or nearly so, to the fluorescence and phosphorescence of benzene and toluene. Under 25-eV excitation there is styong emission a t -500 nm from benzene-doped cyclohexane and thlere are relatively weak bands at 280 and 380 nm. It is quite impossible to determine how much of the latter is due to the very similar emissions of alkanes, but it will be shown that this is not critical. The emissions of alkanes probably arise from fragment ions; i e., they are self-trapped holes,6 and their fate would not be sensitive to addition of 1%benzene. The excitation peaks of Figures 3-6 are those of benzene and toluene luminescence alone for electron energy below -18 e'ii. Those a t or above -18 eV do not a t all resemble the structureless alkane luminescence excitation spectra and these peaks can also be attributed to excitation of ben-
0
I 6
I 8
I I I IO 12 14 Electron Energy, eV
I
I
16
I
Figure 6. Excitation spectrum of 390-nm luminescence for 1 % benzene in nhexane.
zene and toluene emissions. The differences between the two types of spectra can be seen more readily by differentiating the excitation spectra for the 200-nm band of nCloHzz and the 490-nm band of n-CgHls from the preceding article.6 These spectra appear in Figure 7 . The intensities of other bands were too weak for this purpose. The clue to understanding the excitations of Figures 3-6 is provided by comparison with the results from photoelectron spectroscopy in Table I and Figures 3-6. If all of the ionization potentials of cyclohexane and cyclopentane in the low-pressure vapor are decreased by approximately the same amount in the solid, then it should be possible to bring them into register with those luminescence excitations which result from positive-hole trapping, that is from initially delocalized ionic states in the solid. The correlation which has been chosen in Figures 3 and 4 requires a uniform shift of -2.0 eV which can be shown to arise largely from the electronic polarization of the solid for the posiThe Journalof Physical Chemistry, Vol. 78, No. 27, 1974
Timothy Huang and William H . Hamill
2084
T ‘I -i
-0 --I
--2
I
5
I
10 15 20 Electron Energy , eV
I
25
-I-.
’ 1-4
30
Figure 7. Comparison of two types of excitation for 200-nm emission from nCsH.18 and for 390-nm emission from benzene-doped nC6H14.
tive-hole state. The shift for cyclopentane in Figure 5 was also --2.0 eV but the uncertainty is greater due to the limited number of measurements. All luminescence excitation peaks for cyclohexane and cyclopentane systems above -7.5 eV are considered to correlate with energy levels of c-C6H12+ and c-CbHlo+ in the solid. The mechanism for luminescence of the doped systems is then positive-hole formation and migration, trapping by benzene or toluene, charge recombination, and luminescence. The lowest energy of the conducting state of an electron in an insulator, liquid or solid, is commonly represented by VOwhich should not be confused with the zero-point potential energy UO.Referred to a vacuum zero, VOis the negative of the bulk electron affinity, -A. The latter is the least total energy change for (thermal) electron injection and consequently - - A = ‘TO UO.I t is commonly assumed that TO= 0 or that the zero-point kinetic energy of an electron in the conducting state is zero. This is incorrect, possibly by I eV or more.9Jo For a propery zeroed energy scale of a characteristic energy loss spectrum the first (electron injection) peak is located a t A . 3 The zero of energy was arbitrarily located a t the first peak ( A rz 0 for these alkanes in the liquid state”), and Ihe first ionization potential Z, in the solid, e.g., at 7.8 eV in Figure 3, i s correctly described by 1, = (7.8 -t A ) eV. Since Z, is related to the gas-phase ionization potential Z, by d, -a: 1, -. A P h where Ph is the electronic polarization energy of ithe positive hole, i t follows that Ph = 7.8 4- 2A - &. For A ‘v 0,l1Pt, -2.0 eV. The striking featui~eof Figures 3-6 is the relatively wellresolved structure of the ionizing excitations which is not observed optically nor is it observed by energy loss spectra for electron impact, in the gas phase or the solid. It was not observed for the exciitation of the luminescence from neat benzene4 or from many undoped alkane solids using the same apparatus and procedures.6J2 The well-resolved structure will kae attributed to the mechanism of energy mi-
+
+
m e Journal of Physical Chemistry, Vol. 78, No. 21, 1974
gration since this is not required for luminescence from undoped solids. A wide band of vibronic states for each electronic term of alkane ions must be accessible to excitation by electron impact and there should be efficient charge exchange between any of them and a molecule of benzene or toluene. In fact, the half-width of resolved excitation peaks is only -0.8 eV while the electron energy half-width is -0.6 eV. It must be assumed then that many of these vibrationally excited states are not conducting, i.e., that the conducting states lie within a relatively narrow band of energies for each electronic t,erm. This suggests that efficient hole conduction is limited to 0-0, or near 0-0, electron transfers between ionmolecule pairs. The facts also require assuming that vibrationally excited ions do not become conducting as they cool since this would obliterate the selective effect which is observed. By the time an ion has dissipated its vibrational energy, both its configuration and the lattice have somewhat relaxed and the hole should be self-trapped. Electron tunneling to the impurity from nonadjacent donor molecules will not be prompt. At 60°K in solid alkanes there can be no randomized or free charge pairs and the lifetimes of coulombically correlated charge pairs are limited by recombination to sec. Hole migration by hopping probably contributes little. There remains a short-range band-like conduction with nearest-neighbor charge-transfer time of -10-15 sec which cannot couple effectively to vibrationally excited states. (This is an estimated value for cyclohexane which is based upon a band-model calculation for octacosane sec was found.13 Scaling (C28H~8)for which A t s from d = 35.6 A for octacosane for hole migration end to end along the chains to d N 5-.4 for rnigratio across the stacked cycloalkane rings gives At N W 1 5 see.) Consequently, well-resolved ionizations may be selectively detected and higher hole states must retain their electronic identity during migration since electronic relaxation would lead to considerable vibrational excitation. Comparable effects have been observed previously12 in T1+-doped alkali halides above the optical band gap where the mechanism may also be hole trapping followed by recombination. The luminescence excitation spectrum under electron impact was better resolved than the characteristic energy loss spectrum. It is interesting that exciton states of the ionic lattice were also better resolved by the same technique. The distinct excitation peak for benzene phosphorescence from n-hexane a t 6.5 eV in Figure 6 is attributed to energy transfer from a triplet state of n-hexane a t this energy. Excitations of alkanes by low-energy electron impact were observed by Brongersma and Oosterhoff with onsets about 1.0 eV below optical values.l* For cyclopropane and cyclooctane the values were 5.9 and 6.0 eV and they were attributed to excitation of triplet states. Characteristic losses have been observed for several solid alkanes at -6 eV, but only for n-hexane was the energy loss adequately and that determined its choice in this work. There is also a resolved resonance a t 6 eV for chemical decomposition of n-hexane at 77’K under electron impact which was considered to be due to a vibrationally excited triplet state.15 References and Notes (1) The Radiation Laboratory of the University of Notre Dame is operated under contract with the U. S. Atomic Energy Commission. This is AEC Document No. COO-38-944.
Reactions of Ozorie with Ethene and (2) (3) (4) (5) (6) (7) (8) (9)
2085
Propene
K. Hiraoka and W. H. Hamill, J. Chem. fhys., 57, 3870 (1972). K. Hiraoka and W. H. Hamill, J. Chem. fhys., 59, 5749 (1973). K. tiiraoka and W. H. Hamill, J. fhys. Chem., 77, 1616 (1973). W. Wothmari, F. Hirayaima, and S. Lipsky, J. Chem. fhys., 58, 1300 (1973). T. Huang and W. H. Hamill, J. fhys. Chem., 78, 2077 (1974). M. J. Dewar and S. D. Worley, J. Chem. fhys., 50, 668 (1969). F. t-lirayama and S. Lipsky, J. Chem. fhys., 51, 3616 (1969). 0. Klemperer, ”Electron Physics,” Butterworths, London, 1961, p 184.
(10) For solid xenon with A Y 0.6 eV the inner potential is 5.5 eV for 20-eV electron injection. Consequently To 5 eV: A. lgnatiev and T. N. Rhodin, fhys. Rev. B, 8, 893 (1973). (11) R. A. Holroyd and M. Allen, J. Chem. Phys., 54, 5014 (1971). (12) K. Hiraoka and W. H. Hamill, J. Chem. Phys., 58,3686 (1973). (13) W. L. McCubbin, Trans. Faraday SOC.,59, 769 (1963). (14) H. H. Brongersma and L. J. Oosterhoff, Chem. fhys. Lett., 3, 437 (1969). (15) T. Matsushige and W. H. Hamill, J. Phys. Chem., 76, 1255 (1972).
Rate Constants for the Reactions of Ozone with Ethene and Propene, from 235.0 to 362.0 K John T. Herron and Robert E. Huie *I
Institute8for Materials Research, National Bureau of Standards, Washington,D. C. 20234 (Received June 12, 19 74
Publicai’ion costs assisted by the National Bureau of Standards
The rate constants for the reactions of ozone with ethene and propene have been measured over the temperature range 235.0-362.0 K, using a stopped-flow system coupled to a beam-sampling mass spectrometer. The rate constants found, at a total pressure of about 500 N m-2, in the presence of molecular oxygen, were k ( C 2 H 4 ) = (5.42 f 3.19) X lo9 exp(-2557 f 167/T) cm3 mol-l sec-l and h(C3H6) = (3.70 f 1.42) X lo9 exp(--1897 f 109/T) cm3 mol-’ sec-l.
Introduction Although there have been numerous studies of the kinetics of the ozone-olefin reactions a t room temperature (see Stedman, Wu, and Nikil and references therein), there are few data available at other temperatures, the only extensive set of data being that of DeMore2 on C2H2 (243-283 K), C& (178-233 K), and C3H6 (183-193 K). We report here data on the kinetics of the reaction of ozone with C2H4 and C3H6 over the temperature range 235.0-362.0 K. Experimental Sec,t’ion The experiments were carried out using a stopped-flow reactor, shown in Figure 1, coupled to a beam-sampling mass s p e ~ t r o m e t e r . Ozone, ~,~ produced in an ozonizer and stored at -78’ on d i c a gel, was mixed with reactant olefin and carrier gas (AT oir 0 2 ) and was allowed to flow into a 350-cm3 glass reactor. Flow into and out of the reactor was controlled by steinlessi steel solenoid valves, which could be closed simultaneously to entrap the premixed reacting gases. The conterits of the reactor were monitored by means of a beam-sampling mass spectrometer through a 10O-bm diameter orifice. A modulated ion signal was obtained by mechanically chopping the beam, and a phasesensiti,ve detection technique was used. The output of the phase-sensitive amplifier was digitized, sampled at a preselected interval, and stored on a paper tape using a highspeed punch. The reactor was double-walled so that fluid could be circulated to cool or heat the reactor. The temperature of the reactor was measured by means of a copper1
constantan thermocouple inserted into the reactor through a “well” (see Figure 1).The temperature of the circulating heat-exchange liquid was also measured using a copperconstantan thermocouple or an immersion thermometer. The two temperatures were identical under all experimental conditions. The uncertainty in temperature measurement was less than 0.5 K and resulted primarily from potentiometer reading error. The total pressure was about 500 N m-2 and was measured by means of a capacitance manometer. The sensitivity of the mass spectrometer to the olefins was determined for each run using two different gas mixtures of known composition (2-5% olefin in 0 2 ) .
Results For the reaction of ozone with an olefin A 0, + A --+ products
and if [A]
- d[O,l/dt = k*[O,l[Al >> [os],then In [O,] = k,[A]t -t c
(1)
where hl[A] is effectively the first-order rate constant for the decay of ozone in excess olefin. The ozone decay data, in digital form, corrected for any background ion signal a t mle 48, were fitted to eq 1 using a weighted linear leastsquares routine. The weighting function used was W = ( N - B12/(N B ) ,where N is the signal intensity and B is the background correction. The slopes from these fits are the first-order rate constants given in Tables I[ and 11. The stat-
+
The Journal of Physical Chemistry, Vol. 78, No. 21, 7974