Photodissociation Dynamics of Phosphorus ... - ACS Publications

Apr 7, 2010 - (1, 2) However, the photodissociation dynamics of polyatomic molecules is often .... These flight times are required to be reduced to ob...
0 downloads 0 Views 443KB Size
Download PDF
J. Phys. Chem. A 2010, 114, 5271–5278

5271

Photodissociation Dynamics of Phosphorus Trichloride (PCl3) at 235 nm Using Resonance Enhanced Multiphoton Ionization (REMPI) with Time-of-Flight (TOF) Mass Spectrometry Hari P. Upadhyaya,† Ankur Saha,† Awadhesh Kumar,† T. Bandyopadhyay,‡ Prakash D. Naik,*,† and P.N. Bajaj† Radiation & Photochemistry DiVision and Theoretical Chemistry Section, Bhabha Atomic Research Centre, Trombay, Mumbai 400 085, India ReceiVed: January 20, 2010; ReVised Manuscript ReceiVed: March 24, 2010

The photodissociation dynamics of phosphorus trichloride (PCl3) has been studied in a supersonic beam by resonance enhanced multiphoton ionization (REMPI), using time-of-flight (TOF) mass spectrometry. The study is focused on the nascent state of the primary chlorine atom, formed on excitation of the (n, σ*) transition of the molecule around 235 nm. Dissociation of PCl3 and the REMPI detection of chlorine atoms are performed, using the same laser around 235 nm. The photofragments, namely, Cl(2P3/2) and Cl*(2P1/2), are probed, using the 2+1 REMPI scheme in the 234-236 nm region. We have determined the photofragment speed distribution, the recoil anisotropy parameter β, and the spin-orbit branching ratio for chlorine atom elimination channels. Polarization-dependent and state-specific TOF profiles are converted into kinetic energy distributions, using a least-squares fitting method, taking into account the fragment anisotropies. The anisotropy parameters for Cl and Cl* are characterized by values of 0.0 ( 0.05 and 0.20 ( 0.05, respectively. Two components, namely, the fast and the slow, are observed in the speed distribution (P(V)) of Cl and Cl* atoms, formed from different potential energy surfaces. The average translational energies for the Cl and Cl* channels for the fast component are 29.7 and 30.6 kcal/mol, respectively. Similarly, for the slow component, the average translational energies for the Cl and Cl* channels are 9.5 and 9.1 kcal/mol, respectively. The energy partitioning into the translational modes is interpreted with the help of an impulsive model, for the fast component, and a statistical model, for the slow component. Apart from the chlorine atom elimination channel, molecular chlorine (Cl2) elimination is also observed in the photodissociation of PCl3. The observation of the molecular chlorine in the dissociation process and the bimodal translational energy distribution of the chlorine atom clearly indicate the existence of a crossover mechanism from the initially prepared state to the ground state. I. Introduction One of the major goals of a photodissociation experiment is to determine the velocity distributions of the photofragments.1,2 However, the photodissociation dynamics of polyatomic molecules is often complicated because of the involvement of more than one dissociation channel. Though dissociation occurs mainly from an initially prepared state, it can also occur from either a nearby dissociative state, produced by curve-crossing from the initially pumped excited state, or the ground state, produced via an electronic relaxation process, such as internal conversion.3 Atomic eliminations are often observed on repulsive potential energy surfaces, and therefore, these reactions, especially in the photodissociation of halogen-containing organic molecules in the UV region, are well predicted by an impulsive model. The dissociation dynamics involves nonadiabatic crossings, due to strong spin-orbit coupling of the halogen atom, leading to formation of well-defined spin-orbit states of the halogen atomic fragment at the asymptotic limit. Because of these interesting features, halides are considered a prototypical system for studies on direct dissociation dynamics, for investigating the detail nuclear rearrangement during fragmentation.4-6 The photodissociation of various chlorine containing molecules, such as C12, HCl, ICl, PCl3, etc., is interesting because * Author to whom correspondence should be addressed. Electronic mail: [email protected]. † Radiation & Photochemistry Division. ‡ Theoretical Chemistry Section.

of the importance of chlorine atoms in surface etching7,8 and atmospheric science.9,10 It also serves as a well-characterized photochemical source of chlorine atoms, to investigate the mechanism of chlorine atom reaction kinetics.11,12 In addition, small polyatomic precursor molecules are important, since the state distribution of a fragment atom can also, in principle, be determined theoretically, to compare with the experimental results. Recently, there is an interest in the disposal of organophosphorus pesticides and chemical weapons, and consequently, there exists a growing need for information on decomposition and detection of these compounds, particularly phosphorus halides. For example, on-site incineration has been recommended as the safest method for the destruction of these phosphorus halides. However, little is known about their combustion process, and only a few methods are devised for their sensitive detection.13 Phosphorus halides are also important molecules, because these are formed as transient species during the dry halogen etching of III-V semiconductors.14-16 PCln species (where n ) 1-3) are produced, when InP is etched by Cl. Similarly, the etching of both InP and GaP by Br2 yields the corresponding phosphorus bromides.17 The work of Tsang et al.18 has shown that molecules, such as PCl3 and AsCl3, can be used as chemical beam etchants in a semiconductor growth chamber, for etching and regrowth of InP and GaAs, respectively. These PCl3 and AsCl3 molecules contain not only halogen atoms, required for etching semicon-

10.1021/jp100538u  2010 American Chemical Society Published on Web 04/07/2010

5272

J. Phys. Chem. A, Vol. 114, No. 16, 2010

Figure 1. Schematic of REMPI-TOF setup: (1) expansion chamber; (2) ionization chamber; (3) TOF tube; (4) MCP detector; (5) HV supply for MCP; (6) Turbomolecular pumps backed by rotary pumps; (7) gas/ reagent inlet; (8) pulsed valve; (9) skimmer; (10) molecular beam; (11) interaction zone; (12) repeller plate; (13) extraction grid; (14) acceleration grid.

ductors, but also group V elements of the periodic table, making them ideal as chemical beam etchants. Due to less corrosive nature of the group V halides over the conventional etchants, such as HCl,19 these are preferred in chemical beam etching (CBE) processing. Investigations of PCl3 chemical beam etching of InP in a CBE chamber have already shown that excellent etched and regrown surface morphologies can be obtained at high temperatures.18 Thus, knowledge and understanding of the spectroscopy and fragmentation of these phosphorus halides are important.20,21 In literature, there are a limited number of photolysis studies on PCl3.10,22-24 Various other experiments performed on this molecule include He I photoelectron spectroscopy (PES),25,26 threshold photoelectron-photoion coincidence (TPEPICO) spectroscopy and photoionization.20,21 One can also find similar studies on PBr3,20,21 and both experimental27,28 and theoretical29 studies on PF3. In the present work, we have investigated photodissociation dynamics of phosphorus trichloride (PCl3), using resonance enhanced multiphoton ionization-time-of-flight (REMPI-TOF) technique. We have detected both the ground state (2P3/2) and the spin orbit excited state (2P1/2) of chlorine, which lies only 882.4 cm-1 higher in energy. We have also measured the velocity distributions of Cl(2P3/2) and Cl(2P1/2) atoms separately. Another measurable quantity is the electronic branching ratio, defined as distribution of the photofragment Cl over its electronic states, i.e., Cl(2P3/2) and Cl(2P1/2), which was also measured during the course of experiment. These measurements can lead to an understanding of the roles played by the different excited potential energy surfaces. II. Experimental Section The experiment was performed using a molecular beam timeof-flight mass spectrometer system (MB-TOF-MS), consisting of a supersonic beam source and an ionization (beam-laser interaction) region. The system was differentially pumped, using two similar 9 in. turbomolecular pumps (TMP), backed by rotary pumps. An additional 4 in. TMP was used to pump the TOF tube located at the end, near the detector. A schematic diagram of the system is shown in Figure 1. Typical operating pressures in the source and the ionization regions were 5 × 10-5 and 8 × 10-7 Torr, respectively. A pulsed supersonic molecular beam was generated using a solenoid valve, with a 0.8 mm nozzle and 500 µs opening time. The molecular beam was skimmed off through a 1.9 mm diameter conical skimmer to the interaction region. The MB pulse profile and the number density were characterized using a fast ion gauge (FIG) located beyond

Upadhyaya et al. the ion optics, which has a provision for movement along the MB axis. The pulsed valve was located ∼2 cm from the skimmer, and 7.5 cm from the interaction region. The PCl3 sample (99% purity, Aldrich) was used, without further purification. Helium was bubbled through the sample maintained at room temperature, and the mixture was expanded through the nozzle at a stagnation pressure of 1500 Torr of He. It was ensured that any interference to the measurements due to cluster photofragmentation was absent by operating at a low stagnation pressure, and using only the rising part of the molecular beam pulse. The detector system consists of a two-stage WileyMcLaren30 time-of-flight mass spectrometer (TOF-MS), with an extraction and an acceleration regions. The system is mounted vertically, perpendicular to the horizontal MB. The extraction region consists of a repeller electrode (5 cm × 5 cm), which is a solid stainless steel plate, and an extractor grid, mounted 10 mm above the repeller electrode. The acceleration region is defined by the extractor electrode and a grid held at the ground potential, separated from each other by 10 mm. Both these grids (5 cm × 5 cm) are constructed from stainless steel mesh, with 90% transmission. To collect the total ion signal, the extraction region was held at ∼300 V/cm, and the acceleration region was held at ∼3900 V/cm. After passing through the acceleration region, the ion packet passed through a 1035 mm long fieldfree flight tube to the detector. Two deflector plates, placed perpendicular to the detector axis (z axis) allowed the ion packet to be translated in the (x, y) plane to center the ion packet on the detector. The typical field strength for the deflector plates was 2-6 V/cm. The ions were detected by a set of 18 mm dual microchannel plates. A single compact voltage generator, having multiple output voltage ports, was employed to power the TOF ion optics, the deflection plates and the MCP detector. The chlorine atoms were probed, using 2+1 resonance enhanced multiphoton ionization (REMPI) transitions in the region of 234-236 nm. The laser pulses were generated by a dye laser (TDL 90, Quantel), using rhodamine 101 dye. The dye laser was pumped by 532 nm light from the second harmonic of a Nd:YAG laser (YG-981-C, Quantel), operating at 20 Hz. The fundamental dye laser output was frequencydoubled in a KDP crystal and mixed with the fundamental output of the Nd:YAG laser, to obtain an output in the range 230-236 nm. The resultant light had a bandwidth of ∼0.14 cm-1, and was calibrated, using a Fe/Ne optogalvanic (OG) cell and a wavemeter. The above laser output was separated from the rest of laser beams, using four Pellin-Broca prisms. In all the experiments reported in this work, the same laser beam was employed as a pump as well as a probe, i.e., for both photodissociation of the parent molecule and ionization of the photoproducts, Cl(2P3/2) and Cl*(2P1/2) atoms. The laser beam was focused by a lens of 200 mm focal length, and the distance of the lens from the center of the molecular beam axis was varied, to obtain the best ratio of on- and off-resonant signals. The spin-orbit ratio was calculated from the ion intensities for different corresponding transitions, after correction due to their two-photon oscillator strengths. For this purpose, two lines were chosen, one at 42 492.5 cm-1 and another at 42 516.1 cm-1, corresponding to 4p 2D3/2 r 3p 2P3/2 and 4p 2P1/2 r 3p 2P1/2 transitions, respectively. The polarization of the resultant light was rotated, using a double Fresnel rhomb, and a polarizer ensured 99% polarization of the light entering the chamber. The laser power was monitored, using a power meter, and was typically 50-100 µJ/pulse. To obtain the TOF spectrum, the signal was sent to a 500 MHz digital oscilloscope (LeCroy

Photodissociation Dynamics of PCl3 9350A), which was interfaced to a Pentium PC. Subtracting the off-resonant signals from the on-resonant signals effectively removed the minor pump-oil related background contribution to the TOF spectra, and also the contribution from a multiphotonic process. A digital delay/pulse generator, with pulse resolution of 20 ps, was employed as the master to trigger all the instruments for time synchronization. The time delay between the trigger pulse applied to the pulsed valve and the valve opening was obtained by measuring the delay between the trigger pulse and the fast ionization gauge (FIG) signal, employing a digital oscilloscope. This delay is the sum of the time required to open the pulsed valve from its trigger input and that for the molecular pulse to reach FIG from its position of generation, i.e., the nozzle exit. By measuring these time delays for different FIG positions with respect to the skimmer, we estimated the flow velocity of the molecular beam and used it to obtain the time required for the molecular beam to reach the extraction region of the TOFMS. The ion signal was gate integrated by a boxcar, averaged for 30 laser pulses, and fed into an interface (SRS 245), for A/D conversion. A PC was used to control the scan of the dye laser via an RS232 interface and collect data from SRS 245, through a GPIB interface, using a control and data acquisition program. The power dependence measurements were performed by integrating the chlorine atom REMPI spectra, which were obtained by sending the detector output directly to a boxcar integrator, gated on the m/z 35 and m/z 37 peaks in the TOF spectrum. The laser power was varied, and the boxcar output was recorded at each power. The REMPI spectra and survey scans were taken, by recording the boxcar output as a function of the laser wavelength and archiving the spectra on a PC interfaced to the dye laser controller. TOF profiles were taken for three different experimental configuration, Vertical (laser polarization | detection axis), horizontal (laser polarization ⊥ detection axis), and magic angle (laser polarization at 54.7° to detection axis). Doppler broadening of the transitions was always well within the laser bandwidth. III. Analysis We have measured the flight times of Cl and Cl*. These flight times are required to be reduced to obtain photofragment speed distribution, and finally translational energy distribution. To extract this information, a commonly used forward convolution (FC) technique was employed to fit our experimental TOF profiles. There is another approach for this kind of analysis, but it ignores the instrumental broadening factors and is, therefore, valid only for high velocities.31 Deconvolution of the data, prior to direct inversion, corrects this shortcoming, and allows an explicit treatment of the instrumental response function. The instrumental response function was determined by studying the REMPI of aniline beam. The REMPI spectrum of aniline recorded at 34 029 cm-1 (293.77 nm) matches well with that reported in the literature.32,33 The aniline molecular ion signal was measured as a function of laser intensity at the resonance wavelength, and found to be quadratic dependent. This shows that, at 293.77 nm, the REMPI is 1+1 type due to one photon, resonant transition, 1A1 f 1B2, followed by pumping to the ionization continuum by absorption of the second photon. Since the Vz component for the aniline molecule in the beam is almost zero, measurement on a single rovibrational state of aniline molecule can be used to determine the instrument response function. Such measurement showed the instrumental response function to be well described by a fixed Gaussian

J. Phys. Chem. A, Vol. 114, No. 16, 2010 5273 function in the time domain, with fwhm of 27 ns at aniline mass. Under space focusing conditions, this leads to a convolution function in the velocity domain, which is linearly dependent on the extractor voltage Vex. The forward convolution method that we have employed has been discussed elsewhere34,35 and will be only briefly outlined. In a dissociation process of isotropically oriented molecule, using linearly polarized light, the angular distribution of a photofragment is given by36

f(θs,φs) )

σ [1 + βP2(cos θs)] 4π

(1)

where σ is the total photodissociation cross-section, θs and φs are the lab frame polar and azimuthal angles about the axis defined by the laser polarization, β is the anisotropy parameter, and P2 is the second order Legendre polynomial. The β is given by

β ) 2〈P2(cos θm)〉

(2)

where θm is the molecular frame angle between the molecular transition dipole moment and the photofragment recoil direction. In the REMPI-TOF-MS technique, the component of the photofragment that speeds along the TOF-MS axis, which defines the lab frame Z axis, is measured. This speed component results from the averaging of the angular distribution in eq 1 over the photofragment speed distribution g(V) and is given as34,35,37,38

f(Vz, χ) )

[

( )]

1 + βP2(cos χ)P2 ∫|V∞| g(V) 2V z

Vz V

dV

(3)

where VZ is the velocity component along the Z axis, V is the recoil speed of the fragments, β is the anisotropy parameter, P2(cos χ) is the second Legendre polynomial, and cos χ ) εˆ · zˆ, used in the above equation is the projection of the pump laser electric field, εˆ , on the detector axis, zˆ, which is also defined as the angle between the dissociation laser polarization and the Z axis. This procedure differs from “core sampling”, in which an aperture is used to discriminate against the fragments with nonzero velocity components perpendicular to the flight axis (Z-axis). However, in the present work, we have used a procedure of non core sampling data, in which it is assumed that measurement of angular distributions of the chlorine photofragment is independent of the probe polarization. In general, this assumption holds good. But the presence of atomic v · j correlation might make this assumption only approximate. However, these correlations are weak39,40 enough to be neglected. In the case of several contributing channels, eq 3 must be summed over the photofragment speed distribution, gi(V), and anisotropy, βi, of each channel i. The dependence of f(Vz,χ) on β can be eliminated by either measuring the data with the magic angle of χ ) 54.7°, where P2(cos χ) is zero, or coadding normalized profiles with χ ) 90° and χ ) 0°, giving a 2:1 weightage. Both the approaches yield isotropic f(Vz,χ) profiles, and the total center of mass (c.m.) speed distribution is obtained by differentiation of eq 3 at χ ) 54.7° and is given as

g(Vz) ) -2V

d f(V,54.7°)|Vz)V dVz

(4)

5274

J. Phys. Chem. A, Vol. 114, No. 16, 2010

Upadhyaya et al.

The measurements in this work were recorded as TOF spectra, i.e., in the time domain I(t,χ). Under space focusing conditions, a simple linear transformation gives the signal in the velocity domain, I(VZ,χ), according to41

Vz )

qVex(t - t0) m

(5)

where q and m are the charge and mass of the photofragment, Vex is the electric field in the extraction region, and t0 is the mean time-of-flight. The observed velocity domain TOF signal I(VZ,χ) is the result of the convolution of the function f(VZ,χ) with an instrumental response function, which consists of a number of experimental factors, including the duration of the laser pulse, the finite time response of the detector, diffraction from the ion optics grids, and inhomogeneities of the electric fields. Other factors, such as space-charge distortions and the finite dimensions of the ionization region, coupled with deviations from space focusing, may also give rise to a minor contribution. The major task of the analysis procedure is to extract the photofragment speed distribution, gi(V), and anisotropy, βi, of each decay channel i active in the photodissociation, and hence contributing to the experimental TOF profiles I(VZ,χ). In principle, this might be achieved by first deconvoluting the function f(VZ,χ) from I(VZ,χ) using a Fourier transform method. However, this procedure requires an exact knowledge of the convolution function, as well as the use of numerical filters, to suppress the effects of high frequency noise. In the present case, we have used an intrinsically numerically stable forward convolution method. Here, an initial c.m. photofragment speed distribution gi(V) is assumed for each active decay channel i. Application of eq 3 on the magic angle TOF profile I(VZ,54.7°) yields an estimate of the total c.m. speed distribution g(V) and, in turn, some indication of the form of the individual speed distributions gi(V). These are usually modeled with the functional form42,43

gi(Vz) )

(fT)iai(1

- (fT)i)

bi

(6)

where (fT)i is the fraction of the available energy channeled into /Eavail , and ai and bi are adjustable translational modes, Etrans i i parameters. This equation has the advantage of being able to represent many symmetric and asymmetric single-peaked functions, while obeying energy conservation. By taking into account an adjustable anisotropy parameter βi and weight for each decay channel, f(VZ,χ) is simultaneously calculated for the geometries χ ) 0°, 54.7°, and 90°. Convolution with the instrument response function yields simulated TOF profiles, which can be compared with the experimental results. The parameters are then adjusted until a satisfactory agreement with the experimental data is achieved. Once the photofragment speed distributions have been determined, these may be used to obtain the corresponding translational energy distributions. In summary, we have used a forward convolution analysis procedure, which has the capability to include all the contributions active in the photodissociation of PCl3. This procedure provides the weightage of the individual decay processes, as well as their c.m. speed distributions and anisotropy parameters. Translational energy distributions can then be determined from the photofragment speed distributions.

Figure 2. Two +1 REMPI spectra of Cl atom produced in the photo dissociation of PCl3 at 235 nm. The peaks marked are scanned for the measurement of the Cl*/Cl ratio, the intense peak being that of Cl.

IV. Results and Discussion We have investigated the photodissociation of PCl3 at around 235 nm and detected Cl (2P3/2) and Cl* (2P1/2) products via 2+1 REMPI in a one-color experiment using the same laser. A typical 2+1 REMPI spectrum is shown in Figure 2. TOF spectra were measured at resonant wavelengths, from which the off-resonant spectra were subtracted, to eliminate any contribution from MPI processes and contamination from the residual gases. TOF spectra were measured with the laser polarization at three different angles to the detection axis: χ ) 0°, 54.7° and 90°. Analysis of the data yielded the photofragment speed distributions gi(V) and anisotropy parameters βi of the individual channel contributing to the photodissociation. The initial photodissociation process in PCl3 at 235 nm is generation of Cl atom with absorption of one photon.

PCl3 f PCl2 + Cl

∆H ) 76 kcal/mol

(7)

The chlorine atom can also be generated, by subsequent absorption of another photon by the photoproduct PCl2 or Cl2

PCl3 f PCl2 + Cl f PCl + 2Cl

∆H ) 157 kcal/mol (8)

PCl3 f PCl + Cl2 f PCl + 2Cl

∆H ) 157 kcal/mol (9)

The power dependencies of the one-color signals for transitions corresponding to Cl (2P3/2) and Cl* (2P1/2), were performed and a plot for Cl(2P3/2) is shown in Figure 3. The signal is linear in a log-log plot over the range of powers used in the present study. For both types of chlorine atoms, the lines exhibit a slope of 3.1 ( 0.2, which is consistent with one-photon dissociation of PCl3, followed by 2+1 REMPI of the chlorine atoms, assuming that the ionization step is saturated. If we imagine that the Cl atom is also generated by reaction 8 or 9, then in that case the power dependence should be greater than 3, as at least two photons of 235 nm are required for the formation of Cl atoms and another 2 photons are required for REMPI detection. Apart from the power dependence studies, we also systematically measured the shape and the width of TOF profiles of Cl

Photodissociation Dynamics of PCl3

J. Phys. Chem. A, Vol. 114, No. 16, 2010 5275 bandwidth. The measurements were repeated at different laser light intensities, giving similar relative signal intensities. For PCl3 photodissociation, the ion integrated signal intensity ratio, S(Cl*)/S(Cl), has been measured to be 0.83 ( 0.10. From the measured integrated intensity ratio, one can easily obtain the product ratio, using eq 10, in which the relative ionization probability k ) f(Cl)/f(Cl*) was taken to be 0.85 ( 0.10, as reported by Liyanage et al.44 The relative quantum yields, Φ(Cl) and Φ(Cl*), can be determined from the product ratio, and Φ(Cl*) can be expressed as

Φ(Cl*) ) Figure 3. Dependence of the observed Cl(2P3/2) atom REMPI signal from PCl3 photolysis on the laser intensity. The slope of the fitted linear log-log plot is 3.06 ( 0.2.

N(Cl*) N(Cl*) + N(Cl)

(11)

The Φ(Cl*) calculated for PCl3 dissociation is found to be 0.41 ( 0.06. On statistical grounds one expects a quantum yield, Φ(Cl*) ) 0.33. The earlier studies on the determination of the quantum yield of Cl* in the photolysis of the PCl3 molecule at 248 and 193 nm report values of 0.44 ( 0.03 and 0.33 ( 0.03, respectively. These studies were carried out mainly by either using diode lasers tuned to the fine structure transition of the chlorine atom10,22 or performing time-resolved laser magnetic resonance.23 All the studies showed the same value of Φ(Cl*) for a particular wavelength. In the present study, Φ(Cl*) of 0.41 ( 0.06 at 235 nm is much closer to the reported value at 248 nm, which is quite reasonable, since at 248 and 235 nm, the same region of the potential energy surfaces of PCl3 is excited. On excitation of PCl3 at 193 nm, there seems to be either a

Figure 4. Profiles of Cl and Cl* atoms produced in the 235 nm laser photolysis of PCl3 used for the determination of their ratio.

atoms at various laser intensities. In fact, we did see a slight increase in the width of the Cl TOF profile at very high energy. Hence, all the experiments were performed in the intensity range, which are much lower than the intensity at which the change in the shape and width of the TOF profile was observed which ensures that the translational energy distributions and anisotropy parameters are invariant over the laser fluences used. Spin-Orbit Branching Ratio. The relative populations of the chlorine atom fragments in different spin-orbit states are determined by normalizing the integrated intensity, i.e., peak areas (S(X)) of the respective 2+1 REMPI transitions with respect to the laser intensity, and the ratio of the two-photon absorption coefficients. As discussed earlier, the ground Cl(2P3/2) and the spin-orbit excited Cl*(2P1/2) atomic photoproducts were scanned in the region of 235.336 and 235.205 nm, respectively, as shown in Figure 4. The measured area (S(Cl*))/(S(Cl)) of 2+1 REMPI lines is proportional to the product ratio (N(Cl*))/(N(Cl)), with a factor of k,

N(Cl*) S(Cl*) )k N(Cl) S(Cl)

(10)

where N(X) (X ) Cl or Cl*) designates the number density of X produced. The intensity of species X, S(X) is obtained by integrating the measured ion signal intensity over the proper range containing the Doppler width and the probe laser

Figure 5. REMPI-TOF profiles of Cl(2P3/2) produced from the 235 nm photodissociation of PCl3. The circles are the experimental data, and the solid line is a forward convolution fit. The upper, middle, and lower panels correspond to horizontal, magic angle, and vertical experimental geometries, respectively.

5276

J. Phys. Chem. A, Vol. 114, No. 16, 2010

Upadhyaya et al.

Figure 7. Photofragment speed distribution of Cl(2P3/2) determined for the 235 nm dissociation of PCl3. The dashed lines indicate the speed distributions for the fast and slow component for chlorine atom formation channel, respectively; the solid line shows the sum.

Figure 6. REMPI-TOF profiles of Cl(2P1/2) produced from the 235 nm photodissociation of PCl3. The circles are the experimental data, and the solid line is a forward convolution fit. The upper, middle, and lower panels correspond to horizontal, magic angle, and vertical experimental geometries, respectively.

contribution from another higher excited state or a fast internal conversion to the ground electronic state, which alters the quantum yield. Translational Energy Distribution and Anisotropy Parameter. The TOF profiles for the Cl and Cl* atoms, arising from photodissociation of PCl3 at ∼235 nm, were converted to the velocity domain; the representative velocity domain profiles are shown in Figures 5 and 6, as discussed in an earlier section. The three panels shown in Figures 5 and 6 correspond to the TOF profiles recorded for the laser polarization parallel, at the magic angle ∼54.78° and perpendicular to the detection axis, for the Cl and Cl* fragments, respectively. We analyzed the TOF data, using a forward convolution procedure, as described in the earlier section. Here, an initial photofragment speed distribution and the anisotropy parameter β are assumed. The predicted TOF spectra are calculated for the three experimental configurations, convoluted with the instrumental response function determined, as described in an earlier section, and are compared with the experimental results. As atomic v · j correlations are known to be very weak, we can safely assume the TOF profiles to be independent of the probe polarization. The photofragment speed distribution P(V) determined from the data in Figures 5 and 6, for the Cl photofragments, are depicted in Figures 7 and 8. Inspection of Figures 7 and 8 reveals that the P(V) consists of two components. The faster component consists of 50 ( 5% of the total fragments centered at ∼3300 ms-1, while the slower component consists of the remaining 50 ( 5% centered at 1700 ms-1. It was not possible to determine independently the anisotropy parameters for the two components, and hence the anisotropies were assumed to be identical

Figure 8. Photofragment speed distribution of Cl(2P1/2) determined for the 235 nm dissociation of PCl3. The dashed lines indicate the speed distributions for the fast and slow component for chlorine atom formation channel, respectively; the solid line shows the sum.

for each channel and were determined to be 0.0 ( 0.02 and 0.20 ( 0.02, for Cl and Cl*, respectively. The calculated TOF profiles are displayed by the solid line. The partitioning of the available energy into various degrees of freedom of the fragments is mainly governed by the nature of the dissociative potential energy surface. In the case of PCl3, the ultraviolet absorption spectrum shows a broad peak, with a maximum at ∼200 nm, which extends to 250 nm. A detailed MO analysis suggests that excitation at ∼235 nm corresponds to an n f σ* transition on the P-Cl bond, to a repulsive state. It is well-known that the energy partitioning for a dissociative event on a repulsive surface is well described by an impulsive model. So, an impulsive model45,46 has been used in this case, to calculate theoretically the translational energy released to the products. In this impulsive model, the distribution of energy among the product states is governed by the dissociative event, i.e., by the repulsive force acting during the breaking of the parent molecule into the products. This treatment conserves the energy and both the linear and the angular momenta but provides only single-point energies, and not a distribution. The model assumes that all the available energy is stored within the breaking bond. When the bond breaks, the impulse transfers the energy to the photofragments, conserving the linear and the angular momenta. Generally, the impulse is not along the centerof-mass of the separating fragments and thus imparts rotational angular momentum to the departing fragments.

Photodissociation Dynamics of PCl3

J. Phys. Chem. A, Vol. 114, No. 16, 2010 5277

For example, in the present case, PCl3 dissociates into PCl2 + Cl, and the initial recoil momentum of the PCl2 moiety is entirely that of the P atom, Cl2 being at rest. By using only conservation of momentum and energy, and the impulse assumption, one finds that the fraction of the available energy released as translational energy is given by

ET )

( )

µP-Cl E µPCl2-Cl avail

(12)

where µP-Cl is the reduced mass of the P and Cl atoms, µPCl2-Cl is the reduced mass of the PCl2 and Cl, and Eavail is the available energy. In the case of PCl3, the ratio of reduced masses is 0.63. The available energy is given by

Eavail ) Ehν - D00(PCl2-Cl) - ESO

and fT ) ET/Eavail

(13)

where Ehν is the photon energy (122 kcal/mol), D00(PCl2-Cl) is the P-Cl bond dissociation energy (76 kcal/mol), ESO is the spin-orbit energy of chlorine (2.4 kcal/mol), and fT is the fraction of the available energy going into the translational modes of the fragments. Thus, Eavail for the Cl and Cl* channels are 46.0 and 43.6 kcal/mol, respectively. On the basis of the available energies obtained from the above equation, average translational energies of 29.0 and 27.5 kcal/ mol are predicted by the soft-fragment impulsive model at 235 nm, for the Cl and Cl* channels, respectively. The experimental average translational energy for the fast component is found to be 29.7 and 30.6 kcal/mol for Cl and Cl*, respectively, giving fT values of 0.64 and 0.70 for Cl and Cl*, respectively. The experimental calculated translational energies so obtained match very well with the values predicted by the impulsive model. This observation implies that the fast component in the translational energy distribution arises from the repulsive nature of the dissociative potential energy surface. Now, coming to the slow component in the translational energy distribution of the Cl atom, the average energies calculated are found to be 9.5 and 9.1 kcal/mol, for Cl and Cl*, respectively. These values give the same fT of 0.21 for both Cl and Cl*, even though the available energies differ by 2.4 kcal/mol (spin-orbit energy) for these channels. The overall error in the reported translational energy is not more than 10%. Let us now discuss the kinetic energy released for the slow component in the light of a statistical model. A statistical dissociation process occurs predominantly, if the photoexcited parent molecule is so long-lived that the excess energy is partitioned statistically among the available degrees of freedom of the products. This is possible when a rapid internal conversion occurs to the ground electronic state, from which subsequently dissociation takes place. Under these circumstances, a relatively small amount of the excess energy is partitioned into translational motion of the products. An analytical expression demonstrated by Klots,47 relating the mean translational energy release, ET, and the Eavail for a statistical barrierless dissociation process is

Eavail )

(r - 1) ET + ET + 2

hν

∑ exp(hνi/EiT) - 1 i

(14)

where r is the number of rotational degrees of freedom and hνi are the vibrational frequencies of the PCl2 product. In the present case of statistical calculation, the ground state of PCl2 was taken as a bent sructutre, which is confirmed by ab initio calculation, and with the three vibrational modes (200, 480, and 490 cm-1), all having energy significantly less than ET. In this scenario, the last term reduces to ET for each vibrational mode. For PCl2, with r equal to 3, the value ET/Eavail comes out to be 1/5, i.e., 0.20, which matches very well with the experimental value of 0.21. Similarly, other a priori calculations48,49 were also adopted, which also gives a similar fT value of 0.25. This implies that the slow component in the translational energy distribution is mainly due to the Cl atom, which arises from the ground state potential energy surface, after internal conversion via some curve crossing mechanism. This is further confirmed by observation of molecular chlorine (Cl2) ion in the multiphotonic dissociation of PCl3. We clearly observed the Cl2 mass signal in MPI process of PCl3 at 235 nm (∼122 kcal/mol). The power dependence studies of Cl2 ion signal clearly indicate the formation of Cl2 as a primary process. Also, we did not get any Cl2 ion signal in MPI process at 266 nm (∼107 kcal/mol), which shows that the Cl2 elimination channel occurs between 235 and 266 nm. This conclusion was further supported by ab initio calculation, where the transition state for the Cl2 elimination channel from the ground state was determined to be 117 kcal/mol (244 nm) at QCISD(T)/6-311++G(3df,3pd) level of theory. In an earlier study on photolysis of PCl3 at 248 nm, Flynn and co-workers10 measured the translational energy released, using diode laser absorption studies, in which the fine structure transition of the chlorine atom was probed. The translational energy released was determined, using the width of the nascent Cl Doppler profile at zero delay time, which was estimated by an extrapolation of a plot of the width versus delay time. The average ET was determined to be 26 ( 4 kcal/mol, with the fT value of 0.66 ( 0.11. This study was unable to resolve the slow component present, if any, because the width of the nascent Cl atom could not be measured directly; rather it was determined, using an extrapolation method. However, their calculated fT value at 248 nm agrees very well within the validity of the impulsive model, similar to dissociation of PCl3 at 235 nm. V. Conclusion In summary, the PCl3 molecule generates chlorine atom upon excitation at 235 nm, which prepares the molecule in its (n, σ*) state. The nascent distribution of the photofragment chlorine atom is measured by the 2+1 REMPI with TOF mass spectrometry technique. We have determined the photofragment speed distribution, the anisotropy parameter β and the spin-orbit branching ratio for chlorine atom elimination channels, to gain insights into dynamics of chlorine atom formation. Polarizationdependent and state-specific TOF profiles are deconvoluted to get translational energy distributions, using a least-squares fitting method, taking into account the fragment anisotropies. The anisotropy parameters for Cl and Cl* are characterized by values of 0.0 ( 0.05 and 0.20 ( 0.05, respectively. Two contributions, namely, the fast and the slow components are observed in the speed distribution (P(V)) of Cl and Cl* atoms formed in the dissociation process, which emerged from different potential energy surfaces. The average translational energies for the Cl and Cl* channels for the fast component are 29.7 and 30.6 kcal/ mol, respectively. Similarly, for the slow component, the average translational energies for the Cl and Cl* channels are 9.5 and 9.1 kcal/mol, respectively. The energy partitioning into the translational mode is interpreted with the help of an impulsive

5278

J. Phys. Chem. A, Vol. 114, No. 16, 2010

model for the fast component, and a statistical model for the slow component. Apart from the chlorine atom elimination channel, the molecular chlorine (Cl2) elimination is also observed in the photodissociation of PCl3. The observation of molecular chlorine in the dissociation process and the bimodal translational energy distribution of the chlorine atom clearly indicate the existence of a crossover mechanism from the initially prepared state to the ground state. Acknowledgment. We thank Prof. Laurie Butler, University of Chicago, for her immense help and discussion during the course of designing the setup. We also thank Drs. S. K. Sarkar and T. Mukherjee for their constant guidance and keen interest throughout this work. We also acknowledge the initial help during experiments rendered by Mr. Yogesh Indulkar and Ms. Monali Kawade. References and Notes (1) Zare, R. N.; Dagdigian, P. J. Science 1974, 185, 739. (2) Ashfold, M. N. R.; Howe, J. D. Annu. ReV. Phys. Chem. 1994, 45, 57. (3) Satyapal, S.; Johnston, G. W.; Bersohn, R.; Oref, I. J. Chem. Phys. 1990, 93, 6398. (4) Sato, K.; Shihira, Y.; Tsunashima, S.; Umemoto, H.; Takayanagi, T.; Furukawa, K.; Ohno, S.-I. J. Chem. Phys. 1993, 99, 1703. (5) Park, M. S.; Lee, K. W.; Jung, K.-H. J. Chem. Phys. 2001, 114, 10368. (6) Lau, K.-C.; Liu, Y.; Butler, L. J. J. Chem. Phys. 2006, 125, 144312. (7) Ritsko, J. J.; Ho, F.; Hurst, J. Appl. Phys. Lett. 1988, 53, 78. (8) Danner, D. A.; Hurst, D. W. J. Appl. Phys. 1986, 59, 940. (9) Rowland, F. S.; Molina, M. J. ReV. Geophys. Space Phys. 1975, 13, 1. (10) Park, J.; Lee, Y.; Flynn, G. W. Chem. Phys. Lett. 1991, 186, 441. (11) Braithwaite, M.; Leone, S. R. J. Chem. Phys. 1978, 69, 839. (12) Brownsword, R. A.; Kappel, C.; Schmiechen, P.; Upadhyaya, H. P.; Volpp, H.-R. Chem. Phys. Lett. 1998, 289, 241. (13) Belle-Oudry, D. A.; Satyapal, S.; Mussillon, T.; Houston, P. L. Chem. Phys. Lett. 1995, 235, 235. (14) Donnelly, V. M.; Flamm, D. L.; Tu, C. W.; Ibbotson, D. E. J. Electrochem. Soc. 1982, 129, 2533. (15) Takimoto, K.; Ohnaka, K.; Shibata, J. Appl. Phys. Lett. 1989, 54, 1947. (16) Singh, N. K.; Oerlemans, S. Langmuir 1999, 15, 2779. (17) Strubbe, K.; Gomes, W. P. J. Electrochem. Soc. 1993, 140, 3294. (18) Tsang, W. T.; Kapre, R.; Sciortino, P. F. J. Cryst. Growth 1994, 136, 42. (19) Caneau, C.; Bhat, R.; Koza, M.; Hayes, J. R.; Esagui, R. J. Cryst. Growth 1991, 107, 203.

Upadhyaya et al. (20) Boyle, K. J.; Jarvis, G. K.; Tuckett, R. P. J. Chem. Soc., Faraday T rans. 1998, 94, 1045. (21) Boyle, K. J.; Jarvis, G. K.; Tuckett, R. P.; Baumgartel, H.; Jochims, H.-W. J. Chem. Soc., Faraday Trans. 1998, 94, 2073. (22) Tiemann, E.; Kanamori, H.; Hirota, E. J. Chem. Phys. 1988, 88, 2457. (23) Chichinin, A. I. Chem. Phys. Lett. 1993, 209, 459. (24) Bramwell, M. J.; Hughes, C.; Jaeger, S. E.; Whitehead, J. C. Chem. Phys. Lett. 1994, 183, 127. (25) Berkosky, J. L.; Ellison, F. O.; Lee, T. H.; Rabalais, J. W. J. Chem. Phys. 1973, 59, 5342. (26) Cox, P. A.; Evans, S.; Orchard, A. F.; Richardson, N. V.; Roberts, P. J. Faraday Discuss. Chem. Soc. 1972, 54, 26. (27) Reynold, D. J.; Kleef, E. H. v.; Powis, I. J. Chem. Phys. 1991, 95, 8895. (28) Powis, I. J. Chem. Phys. 1993, 99, 3436. (29) Powis, I. J. Chem. Phys. 1995, 103, 5570. (30) Wiley, W. C.; McLaren, I. H. ReV. Sci. Instrum. 1955, 26, 1150. (31) Hwang, H. J.; Griffiths, J.; El-Sayed, M. A. Int. J. Mass Spectrom. Ion Processes 1994, 131, 265. (32) Lubman, D. M.; Jordan, R. M. ReV. Sci. Instrum. 1985, 56, 373. (33) Brophy, J. H.; Rettner, C. T. Chem. Phys. Lett. 1979, 67, 351. (34) McGivern, W. S.; Li, R.; Zou, P.; North, S. W. J. Chem. Phys. 1999, 111, 5771. (35) Bergmann, K.; Carter, R. T.; Hall, G. E.; Huber, J. R. J. Chem. Phys. 1998, 109, 474. (36) Zare, R. N. Angular Momentum; Wiley-Interscience: New York, 1988. (37) Vasudev, R.; Zare, R. N.; Dixon, R. N. J. Chem. Phys. 1984, 80, 4863. (38) Dubs, M.; Bruhlmann, U.; Huber, J. R. J. Chem. Phys. 1986, 84, 3106. (39) Wilson, R. J.; Mueller, J. A.; Houston, P. L. J. Phys. Chem. A 1997, 101, 7593. (40) Bracker, A. S.; Wouters, E. R.; Suits, A. G.; Lee, Y. T.; Vasyutinskii, O. S. Phys. ReV. Lett. 1998, 80, 1626. (41) Syage, J. A. J. Chem. Phys. 1996, 105, 1007. (42) North, S. W.; Marr, A. J.; Furlan, A.; Hall, G. E. J. Phys. Chem. A 1997, 101, 9224. (43) Bracker, A. S.; North, S. W.; Suits, A. G.; Lee, Y. T. J. Chem. Phys. 1998, 109, 7238. (44) Liyanage, R.; Yang, Y. A.; Hashimoto, S.; Gordon, R.; Field, R. W. J. Chem. Phys. 1995, 103, 6811. (45) Tuck, A. F. J. Chem. Soc., Faraday Trans. 2 1977, 73, 689. (46) Riley, S. J.; Wilson, K. R. J. Chem. Soc., Faraday Discuss. 1972, 53, 132. (47) Klots, C. E. J. Chem. Phys. 1973, 58, 5364. (48) Levine, R. D.; Kinsey, J. L. In Atom-Molecule Collision Theorys: A Guide for the Experimentalist; Bernstein, R. B., Ed.; Plenum Press: New York, 1979. (49) Muckermann, J. T. J. Phys. Chem. 1989, 93, 179.

JP100538U