4158
J. Phys. Chem. 1982, 86, 4156-4161
transfer interaction in the excited singlet state does not lead to any enhancement of isc.
Acknowledgment. The present work was supported by a Grant-in-Aid for Special Project Research on Photo-
biology from the Japanese Ministry of Education, Science, and Culture to N.M. We are grateful to Professor K. Nishimoto and Mr. E. Tanaka of Osaka City University for informing us of their results detailing MO calculations before publication.
Multlphoton Ionization of the Trlfluoromethyl Radical Michael 1.Dulgnan,+Jeffrey W. Hudgens,' and Jeffrey R. Wyatt Chemistry Division, Code 61 IO, Naval Research Laboratory, Washington, D.C. 20375 (Received: May 3, 7982; I n Final Form: June 28, 1982)
Resonance-enhancedmultiphoton ionization (REMPI) of gas-phase trifluoromethyl radicals is demonstrated. Four-photon ionization proceeding through a three-photon resonant intermediate state results in a series of multiplets in the 415-4Wnm region. Mass-selective (m/z69) detection and correlation to published vacuum-UV absorption spectra confirm that CF, radicals are the source of most observed structure. Origin of some CF3+ is attributed to resonant multiphoton absorption by parent CF31molecules. Thermal pyrolysis and infrared multiphoton decomposition are compared as sources of radicals suitable for REMPI investigation.
Introduction Resonance-enhanced multiphoton ionization (REMPI) is a sensitive technique for detecting atomic and molecular species.' Transient free radicals are one chemical class for which REMPI provides an excellent new method for detection and spectroscopy; however, to date very few, including NH? and methyl radicalsF4have been examined by this technique. In this paper we report REMPI spectra of trifluoromethyl radicals, CF,, generated by pyrolysis and infrared multiphoton dissociation (IRMPD). New vibronic transitions in CF, are reported. An evaluation of the infrared multiphoton dissociation technique for generating free radicals suitable for REMPI spectroscopy is also given. The trifluoromethyl radical is ubiquitous in such processes as etching of semiconductor surface^,^ halocarbon fire supression,6 and carbon-13 isotope enrichment by infrared multiphoton Kinetic studies may require real-time monitoring of CF3relative concentrations. The method of transient electronic absorption has been used but requires far-UV sources and high radical concentrations often experimentally impractical.lOJ1 CF, has no known fluorescence spectrum so laser-induced fluorescence monitoring is not possible. Current understanding of the electronic structure of CF, is incomplete. The infrared absorption spectrum of CF, has been observed in solid matrix12J3and in the gas phase using a rapid scan technique.14 ESR studied6 conclusively show that the CF, radical is pyramidal in its ground electronic state. The photoionization spedrum16has been recorded in the 86-142-nm region and an adiabatic ionization potential of 9.25 f 0.04 eV was obtained. Analysis of the spectrum, however, was complicated by vibronic structure near the ionization onset. The electronic absorption spectrum in the 146-165-nm region was reported by Basco and Hath0rn.l' Both the photoionization16 and the vacuum-UV absorption" studies report a long band progression with a spacing of -820 cm-'. Since CF3+is expected to be planar, this spacing has been interpreted as the out-of-plane bending frequency (vi) of the ionlawhich is predominantly
'NRC/NRL Postdoctoral Research Associate. 0022-3654182f 2086-4 156$0 1.2510
excited in the conformational change associated with the transition. In the case of the electronic absorption spectrum," this progression has been used as evidence that the observed transitions are Rydberg in nature leaving the CF, ionlike in structure. Robin1' has pointed out that this situation is similar to the n 3s and n 3p Rydberg transitions in NH,, PF,, and PCl, and goes on to suggest that the band system observed by Basco and Hathorn originates in the unpaired electron and terminates in a Rydberg 3p upper state. At least some of the band complexity might then be caused by splitting of the 3p upper state by the ionic core.17 Furthermore, due to the unfavorable Franckxondon factors the origin of the transition has not been observed and may lie 13OOO cm-l below the vertical transition reported to be 64094 cm-l.ll
-
-
-
(1)(a) P. M. Johnson, Ace. Chem. Res., 13, 20 (1980); (b) P. M. Johnson, Appl. Opt., 19, 3920 (1980). (2)J. W. Glownia, S. J. Riley, S. D. Colson, and G. C. Nieman, J . Chem. Phys., 73,4296 (1980). (3)T. G. DiGiuseppe, J. W. Hudgens, and M. C . Lin, Chem. Phys. Lett., 82,267 (1981);J.Phys. Chem., 86,36 (1982). (4)T.G. DiGiuseppe, J. W. Hudgens, and M. C. Lin, J. Chem. Phys., 76,3337 (1982). (5)J. I. Steinfeld, T. G. Anderson, C. Reiser, D. R. Denison, L. D. Hartsough, and J. R. Hollahan, J.Electrochem. Soc., 127,514(1980);D. J. Ehrlich, R. M. Osgood, Jr., and T. F. Deutsch, Appl. Phys. Lett., 36, 698 (1980);C. M. Melliar-Smith and C. J. Mogab in 'Thin Film Processes",J. L. Vossen and W. Kern, Eds., Academic Press, New York, 1978,pp 497-556. (6)R. G. Gann in "HalogenatedFire Suppressants",R. G. Gann, Ed., American Chemical Society, Washington, DC, ACS Symp. Ser., 16, 318-40 (1975). (7) S. Bittenson and P. L. Houston, J. Chem. Phys., 67,4819(1977). (8)M. Gauthier, P. A. Hackett, and C . Willis, Chem. Phys., 45,39 (1980). (9)P.A. Hackett, C. Willis, and M . Gauthier, J. Chem. Phys., 71,2682 (1979). (10)K. Glanzer, M. Maier, and J. Troe, J. Phys. Chem., 84, 1681 (1980). (11)N. Basco and F. G. M. Hathorn, Chem. Phys. Lett., 8,291(1971). (12)D. E. Milligan, M. E. Jacox, and J. J. Comeford,J.Chem. Phys., 44,4058 (1966). (13)D. E. Milligan and M. E. Jacox, J . Chem. Phys., 48,2265(1968). (14)G. A. Carlson and G. C. Pimentel, J.Chem. Phys., 44,4053(1966). (15)R. W.Fessenden and R. H. Schuler, J . Chem. Phys., 43, 2704 ( 1965). (16)C. Lifshitz and W. A. Chupka, J . Chem. Phys., 47,3439 (1967). (17)M.B. Robin, "Higher Excited States of Polyatomic Molecules", Vol. I, Academic Press, New York, 1974,pp 178-91.
0 1982 American Chemical Societv
Multiphoton Ionization of Trlfluoromethyl Radical
Experimental Section The output of the flash-lamp-pumped, Q-switched Nd:YAG laser (Quantel 481C) was frequency tripled to pump a dye oscillator/amplifier (Quantel TDL-111). Dye laser output varied in energy depending on wavelength and dye used (stilbene 420, coumarin 440,450, or 480, Exciton Chemical Co.) but was typically 10-15 mJ at nominal dye wavelength and fell 30-50% -10 nm on either side. Nominal pulse width is 15 ns and bandwidth is 0.5 cm-l. Wavelength accuracy is fO.10 nm.18 Dye laser output was focused by means of a 5-cm fl f/2 quartz lens into the center of the ionizing region of a quadrupole mass spectrometer (Extranuclear). Ions thus generated were mass selected ( m / z 69 for CF,, 127 for I) by the quadrupole and detected by a Channeltron charge multiplier. The resulting signal was current amplified (Keithley 427), digitized, and integrated on a PDP-8E system. Typically 6-12 shots would be averaged and stored, the dye laser stepped 0.02 nm, and the process repeated. An electron impact ion source was regularly employed to tune and calibrate the mass analyzer. Two distinct methods were used to produce trifluoromethyl radicals, thermal decomposition of CF31 (PCR Chemicals), and infrared multiphoton dissociation (IRMPD) of CF31, CF3Br (Matheson), or CF3COCF3(Allied Chemical Division). All gases were purified by several freeze-pump-thaw cycles and purity checked by mass spectrometry. Pyrolysis. The thermal decomposition apparatus has been described in detail el~ewhere.~ Briefly, the sample is allowed to flow through a low-pressure,resistively heated tantalum foil oven directly into the ionization region of the quadrupole mass spectrometer at right angles to the dye beam axis. Pressure in the quadrupole’s chamber under experimental conditions was approximately 10 X lo4 torr. Oven temperature was estimated to be 1000 f 100 “C. Electron-impact production of CF31+ ( m / z 196) showed an approximate 30% decrease with the oven on and it is assumed nearly that same fraction of CF31parent decomposes to CF3 and I. Further evidence is presented in the Results and Discussion section. Infrared Photolysis. IRMPD of CF31,798CF3Br? and hexafluoroa~etone~ is known to proceed at modest fluences producing CF3radicals in high yield. The trifluoromethyl halides were irradiated by a Lumonics 103-2 C02 TEA laser tuned to the R(16) line of the 00°1-0200 transition (1076 cm-l), while the CF3COCF3was irradiated by the R(12) line of the OOol-lOoO transition (971 cm-l). The C02 laser was adapted to run at 2.5 Hz by substituting smaller discharge capacitors (0.15 pF) and replacing the front optic with an 85% reflecting coated germanium output coupler (supplied by Lumonics). Typical pulse energy was -800 mJ and the power/time profile showed the characteristic 200-ns “spike” and 1-ps “tail”.19
-
-
(18) To avoid confusion in comparing multiple-photonand vacuumW single-photon wavelengths, all wavelengths reported here are in vacuo; in the 450-nm region the correction factor for air is approximately -0.1 nm. (19) J. W. Hudgens and J. D. McDonald, J. Chem. Phys., 76, 173 (1982). (20) L. H. Sutcliffe and A. D. Walsh, Trans. Faraday SOC.,57, 873 (1961). (21) E. Grunwald, D. F. Dever, and P. M., Keehn, “Megawatt Infrared Laser Chemistry”, Wiley-Interscience, New York, 1978. (22) M. J. Rossi, J. R. Barker, and D. M. Golden, J. Chem. Phys., 71, 3722 (1979). (23) B. H. Rochey and E. R. Grant, Chem. Phys. Lett., 79,15 (1981). (24) N. Bloembergen and E. Yablonovitch, Phys. Today, 31,23 (1978); A. C. Baldwin and J. R. Barker, J. Chem. Phys., 74,3813 (1981). (25) P. J. Robinson and K. A. Holbrook, “Unimolecular Reactions”, Wiley-Interscience, New York, 1972.
The Journal of Physical Chemistry, Vol. 86, No. 21, 1982 4157
QUADRUPOLE MASS S PECTRO METER
Figure 1. Experimental arrangement for IRMPD production of CF, radicals. The infrared laser fluence within the ionizer could be varied by translation of the germanium lens along the beam axis.
1-1
‘YAG
QSwitch
I
n 10Hz v
i4
2.5 Hz ’
Lamp Sync
-
--lDG1AMP Laser co2 Trig.
AMP
TDG2
25 HZ AMP
I
RFCurrent
Figure 2. Schematic diagram for laser synchronization: (TDG) timedelay generators; (AMP) fast amplifier.
Figure 1 shows the experimental arrangement in the ionization region. The effusive sample beam, C02 laser and dye laser beam meet at right angles. The 25-cm fl f/lO germanium lens is mounted on a translation stage which allows the IR fluence a t the focus of the dye laser to be varied between -3 and 30 J cm-2. The laser synchronization is schematically outlined in Figure 2. A 10-Hz flash lamp sync output generated by the YAG laser is reduced to 2.5 Hz (maximum C02 laser repetition rate) with a digital divide-by-four circuit, delayed, amplified, and used to externally fire the C02 laser. .It was then necessary to open the YAG laser Q-switch at a reproducible delay after the start of the IR pulse. Unfortunately, the jitter between the command to fire and the actual appearance of the IR pulse was quite high ( H O ps). This problem was overcome by noting that there was relatively low jitter (M50 ns) between the electrical discharge in the C02laser cavity and the laser pulse. An rf current probe sensed the discharge and the resulting signal, after delay and amplification, could be used to open the YAG Q-switch. Referring to Figure 2, time-delay generator no. 2 (TDG2) controlled the delay between the IR photolysis pulse and the YAG pumped dye laser probe (0-15 ps). TDG1, on the other hand, (coarsely) controlled the time delay between flashlamp firing and the opening of the YAG Q-switch (-300 ps). YAG laser power was quite insensitive to the TDGl delay over a 30-ps interval near the maximum of the gain curve and so was unaffected by jitter in the firing time of the C02 laser. Focusing properties of the Nd:YAG rods were thus optimally maintained by pumping at 10 Hz while the dye lifetime was maximized by pumping the dye head (26) E. Zamir and R. D. Levine, Chem. Phys. Lett., 67,237 (1979),and references cited within. (27) Aa. S. Sudbo, P. A. Schulz, E. R. Grant, Y. R. Shen, and Y. T. Lee, J . Chem. Phys., 70, 912 (1979). (28) D. E. Powers, J. B. Hopkins, and R. E. Smalley, J. Phjw. Chem., 85, 2711 (1981). (29) C. W. Mathews, Can. J. Phys., 45, 2355 (1967).
4158
Duignan et al.
The Journal of Physical Chemistry, Vol. 86, No. 21, 1982
,
,
,
3-PHOTON FREQUENCY
3-PHOTON FREQUENCY 67700
66gOO
65gOO
,Cml
64100
63270
62460
61640
N
\
E
,
I l l
1
1
I l l 1
7 465 470 475 480 485 nm Figure 4. CF,’ from CF3I and CF,: (a) m l z 69 ion signal from CF,I at room temperature; (b) m l z 69 ion signal observed when CF,I is pyrolyzed in an oven at 1000 OC. Lines labeled Y through 0 denote positions of resonances in CF,I in vacuum-UV absorption experiments’’ at onathird the indicated wavelength. Top trace marks 820-cm-’progression (in three-ph~ton~l) characteristic of CF3 radical.
-
-
141.60,141.16,140.70,140.17, and 139.72 nm (3& = 424.8, 423.5, 422.1, 420.5, 419.2 nm). These vacuum-UV resonances clearly correspond as three-photon resonances to the oven-off features just identified. The relative intensities of the above background bands were small and did not significantly distort the CF, radical REMPI spectrum. A different case, i.e., significant background m / z 69 ion signal due to CF31parent molecules, is portrayed in Figure 4a. Two strong bands at 479.8 and 478.1 nm show intensities nearly equal to the oven-on spectrum (Figure 4b). Sutcliffe and Walsh20 also describe another series of bands in CF,I absorption with their origin at 159.93 nm. Labeled R through Y and located at 159.93, 159.41, 158.89, 158.43, 157.92, 157.60, 157.10, and 155.35 nm, these bands are expressed in the m / z 69 oven-off MPI spectrum in t h r e e - p h ~ t o n . The ~ ~ R and S bands (159.93, 159.41 nm) correlate to the strong m/z 69 MPI resonances of 479.8 and 478.1 nm. Sutcliffe and Walsh20note that these bands are predissociative and lie on a background of continuous absorption. In fact, the R and S bands of CF31at 159.93 and 159.41 nm overlap the 160.11- and 159.26-nm CF, radical absorptions observed by Basco and Hathorn.“ To account for the background structure just described, we considered two separate mechanisms. An element common to both mechanisms is resonant three-photon absorption by CF31 as the rate-limiting step. The first involves REMPI of the parent molecule, followed by photolysis of the parent ion, i.e. CF31+ 3hv
-
CF31+ where the
*
nhv
slow
CF31*
CF3++ I
CF31+
(1)
n = 1, 2, ...
(2) indicates a molecule in a Rydberg state and
(31) Throughout this article the expression “in three-photon” is equivalent to, “at the sum energy of three identical laser photons of the specified wavelength”.
Multiphoton Ionization of Trlfluoromethyl Radical Basco 8 Hathorn
l l l l
I l I I/ I
The Journal of Physical Chemistry, Vol. 86, No. 27, 1982 4159
II II
TABLE I: Observed Transitions in CF, 820 c"
this workd
Basco and Hathornf
this workd
Basco and Hathornf
e e e 6 1 640g 61 970 62460g
60401 60536 61380 6 1 671 b 62457 62 629 62790 63287 63543 63727 64094 64 267 64935 64990 65104
6576v 65900 66580g 66 660 66710 67410g a 67570 67650 682308 68360 690508 69 1 8 0 69860g 70680g 71490g
6 5 720 6 5 902 6 6 569 66 640 6 6 716 67 376 67 513 67 531 67 645 68 222 68 341 c c
a 6 2 760 63270g 6 3 530 63 700 64lOOg 6 4 310 649308 6 5 020 6 5 100
c c C
a Unable to resolve. Not reported by Basco and Hathorn. Not observed by Basco and Hathorn, probably due to interfering absorptions by parent molecules. Attributed by MathewsZ9t o CF, (see text). Reported as three-photon frequency of band head; accuracy approximately r 25 cm-'. e Not observed, probably due to low laser power. Observed in one-photon absorption, ref 11,attributed to CF,. Indicates band heads in 820-cm-' progression,
nicely onto the 820-cm-l progression and are, no doubt, associated with the same electronic transition. Two other smaller bands, located at 433.7 and 484.1 nm (69 180 and 61 970 cm-I), which are not part of the 820-cm-' progression, were seen in this REMPI spectrum. These bands were also not previously reported in one-photon absorption. Features in the CF, REMPI spectrum corresponding to virtually all the one-photon absorption band heads reported by Basco and Hathornll could be discerned, though at poorer resolution. Although it is difficult to determine precisely the temperature of the radicals produced in our experiments, they are certainly hotter than those from which the vacuum-UV absorption spectrum was obtained. Population of low-lying vibrational modes12J3in the ground state and increased rotational temperatures may account, in major part, for the width of our observed CF, bands. Table I lists the observed band heads and those corresponding in one-photon absorption. It should be stressed here that Figure 5 represents a composite REMPI spectrum of CF, assembled from five different scans using four different laser dyes. No attempt was made to normalize for laser power or to substract m/z 69 signal not associated with CF3 radical MPI. For instance, the relative heights of the three shortest wavelength peaks clearly reflect the dye efficiency curve of the stilbene 420 used for the scan. The ionization potential of the CF, radical has been determined by the photoionization technique to be 9.25 eV (54 600 cm-').16 This ionization limit corresponds to an energy equivalent of four 536-nm photons. The three-photon resonance nature of the observed transitions has been clearly established. Therefore, the REMPI spectrum in Figure 5 represents a 3 + 1 process. That is, three dye laser photons excite the radical to a resonant level and a fourth single photon ionizes it. REMPI tends to favor the observation of Rydberg states over shorterlived valency transitions.lb In this view, the observation of MPI resonances listed in Table I strengthens the suggestion by Basco and Hathornl' and Robin17that these are, in fact, Rydberg transitions. However, the congestion in the spectrum and the lack of further members of the
4160
The Journal of Physical ChemMry, Vol. 86, No. 21, 1982
N
\
E
PYROLYSIS
nm DYE LASER WAVELENGTH 453
455
457
Figure 6. Comparison spectra of CF, radicals produced from CF,I by (a) thermai pyrolysis and (b) I R mutlphoton photolysis. Dotted line at 454.3 nm denotes feature attributed to a hot band. See text.
Rydberg series make a more detailed assignment difficult. In analyzing the vacuum-UV absorption spectrum that they obtained for CF,, Basco and Hathorn” noticed a correlation of 17 band heads with those reported by Mathews29which he had tentatively attributed to difluorocarbene, CF2. Basco and Hathorn concluded that the Mathews assignments were in error and that CF, was the source of the absorption. We too see those same coincidences with the absorptions reported by Mathews,29plus an additional five not observed by Basco and Hathornl’ in the REMPI spectrum of CF,. Bands reported by Mathews at 69018,69 156,69832,70681,and 71 495 cm-l are seen in our MPI spectra at 69 050,69 180,69 860,70 680, and 71 490 cm-l in t h r e e - p h ~ t o n .The ~ ~ fact that only ions of m/z 69 (CF,+) are detected for our spectra unambiguously demonstates that CF, (mlz 50) cannot be their origin. Infrared Multiphoton Production of CF,. Infrared multiphoton decomposition (IRMPD) was also employed to produce trifluoromethyl radicals. It is well-known that IRMPD of halogenated trifluoromethyl compounds7v8or hexafluor~acetone~ yields CF3 efficiently. The versatility of IRMPD for the generation of a wide variety of free radicals21J2as well as a choice of precursors for each radical is very appealing. The high concentrations of IRMPDgenerated radicals could be probed by the ionizing laser(4 before recombination and diffusion depleted their population. Recently, it has been shown that REMPI may be a useful probe for detection of IRMPD products.23 REMPI of trifluoromethyl radicals produced by IRMPD was attempted with mixed results. Figure 6, which compares IRMPD and thermal pyrolysis of CF,I as a source for CF,, is illustrative. Several differences are at once apparent. First, a new band centered at 454.3 nm (66 040 cm-l in three-photon) appears in the REMPI spectrum of the C02-laser-producedCF,. This new band coincides with a minor (but reproducible) feature in the pyrolytically produced CF, spectrum that might otherwise have been missed. Second, MPI of the IRMPD-generated radicals results in a broader spectrum. Third, signal to noise is
Duignan et ai.
somewhat poorer for the IR photolysis case. Increasing the delay between photolysis and dye laser pulses from 1.5 (as in Figure 6) to 10 MS decreased the average m/z 69 ion current, perhaps narrowed the bands slightly, but also resulted in even less favorable signal to noise. Decreasing that same delay increased the average ion current but broadened the bands even further, washing out most structure. The ion current strength and widths of the CF, bands also increased with increasing COz laser intensity. Tuning the mass analyzer to m/z 127 (I+)helped to locate strong, iodine atom resonances that disappeared when the C02 laser was blocked, indicating that trifluoromethyl iodide efficiently decomposed. Using CF,Br or (CF3)C0as an IRMPD source of CF, resulted in a high m/z 69 MPI ion current, but with essentially no wavelength-dependent structure. The above observations are reconciled by recognizing that IRMPD is both a dynamid and a statistical process.% At high infrared intensities molecules can be pumped well above the decomposition threshold. The decomposition rates of precursors and the average energy available for distribution among the fragments can be computed with RRKM theory.25 The actual internal energy distribution of the fragments may be estimated by statistical phase space theory, but exit-channel effects can alter these distributions.26 If either the rotational or vibrational temperature of the CF, fragments is elevated enough, the already congested CF, spectrum could “wash out” through hot-band formation. In general, the faster the infrared up-pumping rate, the hotter the average precursor molecule becomes before dissociating. Hotter precursors produce hotter fragments. Infrared laser intensity influences the product energy since a higher intensity increases the up-pumping rate. Evidence and theory also predict greater energy for distribution between fragments as the reaction barrier for decomposition increases.27 We note that only CFJ which possesses the lowest decomposition barrier among the precursors of this study yielded any m / z 69 REMPI structure. Measurements of translational energy distributions of fragments produced by IRMPD in beamsz7 imply that, under collisionless conditions, high rotational temperatures would not be expected. Vibrational “hotband” absorption is therefore seen as the more likely cause of the peak broadening observed under these conditions. In our view pyrolysis produced radicals with lower vibronic temperatures than those produced by IRMPD. Incorporation of a supersonic jet expansion which would cool the nascent radicals offers the next logical improvement to radical production by IRMPD.28 The disappointing signal to noise in the IRMPD experiments then probably resulted from three factors. First, since the parents and radicals were hotter than those pyrolytically produced, peak/base line contrast is less favorable than a thermal source. Second, the C02 laser, pushed to operate at 2.5 Hz,produced pulses that varied shot-to-shot in energy (on the order of 10%) and beams that were spatially inhomogeneous. The MPI probe beam then sampled different concentrations and temperatures of CF, radicals with each shot. Last and probably least important, synchronizationjitter between lasers may have caused number density and nascent radical temperature effects similar to laser shot-to-shot fluctuations. Conclusions It is demonstrated that trifluoromethyl radicals undergo four-photon ionization in the 415-490-nm region via three-photon resonance with a molecular Rydberg state. Extensive excitation of an 820-cm-l umbrella mode in the upper state results from a pyramid-to-planar conforma-
J. Phys. Chem. 1982, 86, 4161-4164
tional change associated with the transition. Several previously unobserved vibronic transitions are reported. Infrared multiphoton dissociation (IRMPD) of several CF3-containing compounds resulted in hot CF3 radicals. Under collisionless conditions, this may limit IRMPD's usefulness as a radical source (especially for larger radicals)
4101
suitable for spectroscopic investigation. REMPI of trifluoromethyl radicals is shown to be a sensitive, real-time technique for their detection, while the high number densities, long path lengths, and other experimental difficulties associated with alternate detection methods (IR or vacuum-UV absorption) are avoided.
Photolytic Decomposition of Sodium Tetrahydroaluminate Powdert D. Doughertyt and P. J. Herley' Department of Materleis Science and Engineering, State University of New York, Stony Brook, New York 11794 (Received October 12, 1981; In Finel Form: June 28, 1982)
The photodecomposition of NaAlH4 has been investigated at room temperature as both pristine and %o y-ray irradiated powder. The photolytic rate vs. time curves show an initial acceleration to a maximum rate followed by a deceleratoryperiod which merges into a linear rate. The linear rate approximates a second-orderdependence on light intensity in both the pristine and irradiated (1.0 X lo6 rd) powder. Fitted rate curves can be resolved into a linear term, one long-lived exponential, and at least one short-lived exponential, depending upon the photolytic light intensity and y-ray irradiation dose. The long-lived exponential rate constant is independent of light intensity and/or y-ray irradiation dose. The isothermal fractional decomposition vs. time curves are sigmoidal and preceeded by a small initial evolution of gas. y-ray irradiation accelerates the decomposition reaction. The data for pristine and irradiated powder are best fitted by Avrami-Erofeyev nucleation and growth kinetics. Simultaneous UV irradiation and thermal decomposition produces complex decomposition curves below the melting point. The isolated photolytic and thermal processes can be well represented analytically and can be interpreted by conventional phenomenological models. The coirradiated data points to a synergistic (nonadditive) decomposition process.
Introduction This investigation extends the series of thermal and photolytic decomposition kinetic studies on alane'-, and the alkali-metal alanatea" to NaAlH, powder. It includes the effects of simultaneous UV photolysis and thermal decomposition below the reported melting point8 of 183 "C. The available literature does not contain references to the photodecomposition of NaAlH,, but reportedlyB NaAlH4 powder (white) does not become grey under the action of (visible) light, as does LiA1H4. Our approach to photodecomposition envisages that the decomposition results from interactions between photonproduced energy carriers and defects in the solid. A phenomenological rate expressionlo derived from these assumptions relates the decomposition rate, Rr(t),at constant lamp intensity, I, to the time, t:
-
n
Rr(t) = f i
+ Cfi exp(d;t) i=2
(1)
All of the constants, f i and di,represent various mechanistic combinations of the rates of formation or destruction of singly excited sites. A further theoretical outcome relates a steady-state (linear) photolysis rate, R,,, (obtained at long times), to the relative light intensity, I/Io: &/r,,) = t1 PU/Io)I (2) where a and p are constants. The purpose of this work is to interpret the photodecomposition data of NaAlH, by examining the effects of intensity, temperature, and y-ray preirradiation.
*
Present address: c/o Nuclear Energy Dept., Waste Management Division, Brookhaven National Laboratory, Upton, NY 11973. 0022-365418212086-4161$01.25/0
Experimental Section Materials. Commercial N&H4 powder (Alfa Products, 95% purity) was recrystallized'l at 50 "C from saturated tetrahydrofuran (THF) solution by evaporation under dry nitrogen. The resulting crystallites were rinsed once with dry THF and ground coarsely in a Pyrex mortar. The powder which passed through a U.S.A Standard Testing Sieve No. 230 (mesh size 63 pm) was dried in a vacuum oven at -0.01 mmHg and 50 "C for 12 h and stored over P,O+ Metals other than Na and A1 were not detected by SEM-EDAX analysis of the microcrystallites. Apparatus and Procedure. The sample was decomposed in the Pyrex-glass system described previ~usly.~ Samples (100 mg) were spread evenly in a 1.2-cm diameter fused-silica boat and pumped overnight in the decomposition chamber at room temperature to minimize any residual outgassing. Before decomposition both the lamp and the furnace were allowed to stabilize for 1 h. Then (1)P. J. Herley and R. H. Irwin, J. Phys. Chem. Solids, 39, 1013 (1978). (2)P. J. Herley, 0. Christofferson, and R. H. Irwin, J. Phys. Chem., 85,1874 (1981). (3)P.J. Herley and 0. Christofferson,J.Phys. Chem., 85,1882(1981). (4)P.J. Herley and 0. Christofferson,J.Phys. Chem., 85,1887(1981). London, Ser. A, (5)W.E.Garner and E. W. Haycock, Proc. R. SOC. 211,335 (1952). (6) P. J. Herley and D. A. Schaeffer, J. Phys. Chem., 82,155 (1978). (7)P.J. Herley and D. H. Spencer, J. Phys. Chem., 83,1701 (1979). (8)J. A. Dilta and E. G. Ashby, Inorg. Chem., 11, 1230 (1972). (9)H. Clasen, Angew. Chem., 73, 322 (1961). (10)P.W.Levy and P. J. Herley, React. Solids, Int. Symp., 6th, 75 (1968). \----,-
(11)J. W.Lauher, D. Dougherty, and P. J. Herley, Acta Crystallog., Ser. B, 35,1454 (1979).
0 1982 American Chemical Society