The 1+1 resonant multiphoton ionization of hydrogen iodide through

Jul 12, 1993 - room-temperature methyl iodide with a tunable dye laser near the threshold for ... Dilute samples of hydrogen iodide (Matheson, 98.0% p...
1 downloads 0 Views 846KB Size
J. Phys. Chem. 1993,97, 13508-13514

13508

1+1 Resonant Multiphoton Ionization of HI through the First UV Continuum Mark A. Young Department of Chemistry, University of Iowa, Iowa City, Iowa 52242 Received: July 12, 1993; In Final Form: October 14, 1993'

We have examined resonant multiphoton ionization (REMPI) in HI using a tunable nanosecond dye laser. The parent ion, HI+, is produced by a 1+1 REMPI process with the excitation wavelength resonant with the first UV continuum of HI. Sharp features in the wavelength-resolved spectrum of HI+ are recorded and are assigned to two-photon transitions to autoionizing Rydberg states with a finite lifetime. Differences in the two-photon spectra compared to data obtained by single-photon V U V photoionization are noted. A simple kinetic analysis of the data is employed to illuminate the mechanism of ionization and to explain the observed laser flux dependence of the parent ion signal.

The application of powerful, tunable pulsed laser sources to induce multiphoton ionization (MPI) in atomic and molecular species has become a common, yet potent, spectroscopic probe of photochemical and photophysical processes. Resonance enhancement of multiphoton absorption results when the energy of one, or more, of the incident photons matches that of an intermediatelevel in the system under study. In molecular species, an intermediate state of strongly predissociative character may be accessed and subsequent ionization must compete with intramolecular decay. An interesting situation is encounteredin a molecule if the intermediate state is purely repulsive in nature, engendering rapid dissociation. We have examined such a process in the hydrogen halide, HI, by employing a 1+1 MPI process that is one-photon resonant with the first ultraviolet continuum. The structureless continuum absorption, due to a n u* transition, begins near 300 nm and has a maximum at 220 nm. Despite the fast photodissociation known to occur, we are still able to observe the parent ion, HI+,using conventionalnanosecond laser sources. An analogous situation has been addressed by various investigations of MPI in the polyatomic species, CH31. T_heelectronic structure of CH31is similar to HI and the excited A state leads to a broad, continuous absorption beginning at approximately 320 nm and peaking near 260 nm. Absorption into the A state continuum at 292 nm results in dissociation of the C-I bond on a timescaleof = O S ps.' Bemsteinandco-workers2studiedmethyl iodide from an effusive caplllary source using a 2+1 REMPI process resonant with the A state continuum. Mass specific detection was afforded by a time-of-flight (TOF) mass spectrometer. Excitationwith several millijoules of 266-nmradiation (the fourth harmonic of a Nd:YAG laser with a pulse width of = 5 ns) yielded no parent ion, CH31+,while fragment ions, I+ and CH3+, were readily observed. Danon et aL3excited a sample of CD31prepared by expansion of the neat gas from a narrow bore capillary positioned in the ionizationregion of a quadrupole mass filter. Their Nd:YAG laser, also operating at 266 nm, reached maximum intensity levels of lo7 W/cmZ. In contrast to the previous results, CD3I+ was observed with a quadratic intensity dependence which decreased as higher intensities were used. It was proposed that the CDsI+ originated from a two-photon ionization process in vibrationally hot speciespresent in the sample. Vaida and co-workers utilized direct absorption4 and MPI probesSof supersonicexpansions of neat methyl iodide to identify the presence of dimers and higher order cluster species, which appeared to be formed in appreciable concentrations. Under

-

Abstract published in Adounce ACS Absrrucrs, December 1, 1993.

conditions that favored the formation of bare, monomeric species, excitation with excimer laser radiation at 308 nm (2+1 REMPI process) or 248 nm (1+1 REMPI process) failed to produce detectable traces of the parent ion.s Intensities estimated at 100 MW/cmZ were produced by the laser system used. Alteration of the expansion conditions to produce dimers and larger clusters, with similar photolysis conditions, resulted in the detection of CH3I+ as well as other ions corresponding to cluster reaction products. The formation of the parent ion was thought to be facilitated by cluster-inducedcaging which inhibits the dissociation of the initially excited CH31molecule. These researchers ventured that the discrepancy in the previous work was due to the sample conditions employed by Danon et aL3which may have promoted the formation of clustered species. In related experiments, Gedanken et aL6 examined MPI in room-temperature methyl iodide with a tunable dye lasfr near the threshold for single-photon resonance with the A state continuum. While monitoring the total ion current, a sudden transition from broad absorptions,characteristic of 2+ 1 REMPI resonant with the 6p Rydberg state, to very sharp features, assigned to the I atom fragments of CH31photodissociation, was recorded. The 4+2 REMPI process was also investigated by irradiating with the dye laser fundamental and monitoring products with a TOFmass spectrometer. Several prominent features in the CH3I+ mass channel were detected. These results were rationalized by noting that the laser fundamental can access the A2 component of the A state in a two-photon process, which is forbidden in a single-photon transition. The AZcurve may be slightly bound with a sufficiently long lifetime to permit subsequent photon absorption and, hence, ionization. Such a conclusion is justified by the results of CH31photolysis in rare gas mat rice^,^ indicating that the Az state is bound by approximately 1400 cm-I. More recently, zero-electron-kinetic-energy (ZEKE) spectroscopy has been utilized toobserve vibrational structure in CH$ produced froma 1 + 1 MPI process resonant with thecontin~um.**~ These experimentsmade use of nanosecond excimer-pumped dye laser sources to generate the parent ion. No evidence of cluster formation was observed. The photoelectron spectrum reveals an extended progression in the C-I stretching mode which can be understood as the influence of the dissociative intermediatestate. It has been convincingly demonstrated that multiphoton ionization with ultrafast laser sources can effectively compete with molecular dissociation. Szaflarski and El-Sayed compared MPI in CH31with nanosecond and picosecond laser sources using TOF detection.lO Utilizing the fourth harmonic of a nanosecond Nd:YAG laser, they were unable to observe CH3I+ in roomtemperature samples, However, the output of a mode-locked Nd:YAG laser at the same wavelength, but with a pulse width of

0022-3654/93/209113508$04.00/0 Q 1993 American Chemical Society

1+1 Resonant Multiphoton Ionization

The Journal of Physical Chemistry, Vol. 97, No. 51, 1993 13509

m+ r+

1'

1

I

A

B

A

h

Y

E

3 I* TIME OF FLIGHT

Figure2. Mass-resolved photoionization spectra of an expansionof 0.2% HI in He at a stagnation pressure of 1 psig. The excitation wavelength was 240.0 nm in A and 238.4 nm in B.

R(h Figure 1. Potential energy curves for HI. The arrow represents the

energy of the photon in the 1+1 REMPI process at the threshold for production of HI'. The repulsive curves are taken from ref 18.

30 ps (-2000 MW/cm2), produced abundant parent ion signal, as well as smaller fragments. The REMPI process at 266 nm requires three photons to yield CH31ions with a total energy of 13.98 eV, well above the IP of 9.54 eV." Excitation with picosecond pulses at 246 nm, generated via Raman scattering (5 pJ pulse energy), permitted a 1+1 REMPI processjust above the IP which resulted in the production of parent ion only, suppressing fragmentation. Other investigators have reported the use of ultrafast laser sources to observe CHsI+ from the monomer and from clusters.lJ2 In the current experimental effort, the hydrogen halide, HI, was studied using a 1+1 REMPI process that is resonant with the first UV continuum. A diagram of the relevant potential energy curves is shown in Figure 1. The accessible states that comprise the continuum are known to be repulsive, and thus, the situation is similar to that discussed above for CH31. Using conventional nanosecond dye laser sources, we have been able to observe the parent ion, HI+,in dilute, supersonic expansionsunder conditions which discourage the formation of dimers and higher order clusters. In addition, wavelength-resolved studies reveal sharp, well resolved structure that appears at the expected threshold for HI+ formation. A previous investigation by Codling et a1.I3J4detected the parent ion as well as multiply ionizedspecies, HIn+, n > 1. However, that work involved the use of a highintensity, femtosecond light source with fluxes on the order of photons/scm*. The resultant electric fields (a10 V/A) were sufficientlylarge to result in field-induced ionization. Our studies involvefluxes that are 6-7 orders of magnitude less and are unlikely to involve similar ionization processes. Experimental Section Dilute samples of hydrogen iodide (Matheson, 98.0% purity) and helium carrier gas were prepared in a passivated stainless steel sample cylinder and allowed to mix for several days. The helium gas was passed through a molecular sieve filter prior to mixing in order to remove any residual moisture. All gas sample lines were well passivated with HI prior to use. The spectra presented here were recorded with samples having a measured concentration of 0.2% HI in helium.

The gaseous sample was expanded at pressures of 1-40 p i g through a commercial pulsed valve (General Valve) with a pulse width of =200 ps. The supersonic jet source was coupled to a TOFmassspectrometer (R.M. Jordan) equippedwitha reflectron. While the reflectron configuration offers an increase in the mass resolution, it was not a critical advantage for the work described here. The supersonicexpansionwas skimmed by a 1"diameter conical skimmer prior to entering the laser interaction region. A digital delay generator allowed the temporal delay between the opening of the valve and the triggering of the laser pulse to be precisely controlled. The laser source was an excimer pumped dye laser combination (Lambda-Physik EMG 200/FL3002) using C480 and C460 dyes and a BBO crystal to yield radiation tunable over the 255-225 nm region with an approximate UV bandwidth of 10.4 cm-I. A Pellin-Brocca prism arrangement was used to separate residual fundamental light from the second harmonic output. The maximum UV pulse energy utilized was approximately 1 mJ/ pulse. The dye laser output was attenuated for intensity dependencestudies by partially blocking the excimer pump beam into the final dye amplifier cuvette. In estimating the amount of energy actually impinging on the sample molecules, the attenuation of the energy by the beam steering prisms, lens, and windows was measured with a pyroelectric energy meter. A 15cm focal length planoconvex lens was generally used to focus the dye laser beam into the interaction region of the TOF, although for somestudiesa longer focallength lens of 70cm wassubstituted. The focal lengths are calculated at the UV wavelengthsof interest. The dye laser fundamental was calibrated by recording the response of a neon hollow-cathodelamp over the wavelength range used in our study. The measured deviations were fit to a linear function, and the resultant wavelength accuracy was about 0.2 cm-I. Photoionswere detected with a dual microchannel plate (MCP) detector. Mass-resolved scans were recorded with a digital oscilloscope (LeCroy 9410,100 MHzA/D rate), and wavelengthresolved spectra employed a boxcar integrator to selectively monitor the signal due to the desired mass. The assignment of mass peaks was confirmed from the arrival time of benzene ions, recorded under similar operating conditions. ReSultS

In Figure 2, mass-resolved scans resulting from the irradiation of a dilute (0.2% HI in helium) expansion at a total stagnation pressure of 1 psig are shown. At a wavelength of 240.0 nm, only I+ ions, m / z = 127 amu, were detected. A small change in the wavelength to 238.4 nm manifested a second, heavier mass signal that corresponds to m / z = 128 amu, the mass of HI+. No other mass signals were detected under these conditions. At higher

Young

13510 The Journal of Physical Chemistry, Vol. 97, No. 51, 1993 n=S

1

6 I

sakr3

I

sakr4.5

I

I

3

83Mo

83700

84100

84WO

84500

8

I I

I I

I

I

I

I

I

I

86ooo

86oM

86100

861%

86200

WAVENUMBERS (em") Fi-4. Expanded view of a portion of the HI spectrumshowing features in series 3-5 with n = 6 . The peaks are labeled with the notation described in ref 29.

I

I

I

I

89%

85700

10

9 I

I

6

85300

I

85900

I

I

I

1

I

I

*

85700

*

'

I

86100

.

"

I

86500

"

.

I

86900

"

'

I

~

87300

~

~

87700

I

~

'

'

l

'

88100

WAVENUMBERS (cm-l) Figure 3. Extended wavelength-resolvedHI photoionization spectrum.

Features assigned to spin-orbitautoionizing Rydberg series are identified according to the series labels of ref 28.

stagnation pressures of 40 psig, a signal at m / z = 254 amu, corresponding to Iz+, was detected and was indicative of cluster chemistry due to the formation of larger clusters. The Iz+signal could be enhanced by adjusting the pulsed valve delay such that the laser sampled later segmentsof the gas expansion where larger clusters are known to be formed. Adjustment of the valve delay at stagnation pressures of 1 psig resulted in no changes in the observed mass spectrum, or the observed wavelength-resolved spectrum (vide infra). However, the data presented here for 1 psig stagnation pressure were recorded with a valve timing such that early portions of the expansion, where monomeric species are expected to dominate, were interrogated. The boxcar gate was set to integrate the HI+ signal and the wavelength of the dye laser scanned to produce the extended spectrum contained in Figure 3, plotted as a function of the twophoton energy. Sharp, well resolved features were observed overlapping a weaker continuous signal, and the onset of the spectrum, -239 nm, was very near the two-photon energy threshold for the production of HI+ (IP = 10.386 e V ) . Several of the features were very narrow with spectral widths as small as 1.7 cm-I, close to the limiting resolution of the ionizing laser which has a spectral bandwidth of 0.8 cm-I at the two-photon level. The HI+ mass signal decreased significantly at higher energies until eventually,beyond -88 700 cm-I, no distinct signal was observed. A more detailed view of a portion of the spectrum is offered in Figure 4. The threshold of the spectrum was found to be a sensitive function of the applied extraction voltage of the TOF, shifting

I 63200

I

.

836rm

'

.

I

'

84OOo

.

"

.

"

84400

I

"

84800

WAVENUMBERS (cm-1) Figure 5. Wavelength-resolvedspectra of the HI+mass channel plotted

as a function of the two-photon energy. The spectra displayed were collected at different settings of the extracting electric field in the TOF, aslabeled. Thearrowsindicatcthepositionofthethresholdforappcarance of the HI+ signal. Features marked with an asterisk are due to contamination from the strong, nearby I+ signal. to lower energies as the electric field is increased (Figure 5 ) . An external electric field can reduce the measured IP of molecular species by field-ionizing highly excited Rydberg levels.I6 A classical model for the IP ~ h i f t , ~AIP, ~ , I yields ~

where E is the field strength in V/cm. The observed threshold of the HI+ signal is plotted as a function of the square root of the extracting electric field in Figure 6. The data are nicely fit by a straight line with a slope of =6, the expected value for an adiabatic field ionization mechanism. The zero-field threshold

1+ 1 Resonant Multiphoton Ionization

The Journal of Physical Chemistry, Vol. 97, No. 51, 1993 13511

3 2

Y

:

83400

!

I

I

0

10

m

I

I

I

3 0 4 0

50

v‘E (Voitr/cm)”’ Figure 6. Plot of the threshold for appearance of the HI+ signal versus the square root of the extraction field. The experimental points can be fit by a straight line with a slope of -6 cm-l (V/cm)-lI2,as predicted by the model for field-induced ionization.

1

n\

.

834GQ

83600

83800

OOkM

0.1-1.0 MW/cm*

84OOo

WAYENUMBERS (em-*)

Figure7.Wavelengt-resolved spectra of HI+ for several different focusing arrangements of the incident laser beam, as labeled. Rough estimates of the maximum intensities achieved for the different conditions are also given in the figure.

is calculated to be 83 766 cm-I, very near the accepted IP of HI (83 771 cm-I). Routine operating conditions entailed a 2000-V extraction voltage applied over a distance of 1 inch, or 787.4 V/cm. We have also examined the spectrum with a longer focal length lens as well as with the unfocused output of the dye laser, resulting in a larger spot size in the interaction region and a consequent reduction of the laser flux impinging on the jet-cooled sample. The results are shown in Figure 7. There was a definite decrease in the observable ion signal as the flux was decreased, but the largest peak was still distinct in the case of the unfocused laser, albeit with a much reduced signal-to-noiseratio. However, there were no obvious shifts or broadening of the spectral features that would indicate nonlinear laser-induced effects. The intensity dependence of the HI+ signal was recorded, shown in Figure 8,

Figure& IntensitydependenceoftheHI+featureat83 880cm-I. Overall, the data are fit reasonably well by a linear dependence, as indicated by the dark line. There is noticeable curvature to the data, illustrated by the line connecting the experimental points. The maximum intensity applied is approximately 8&160 MW/cm2, corresponding to a photon flux of 1-2 X photons/s.cm2.

with the 15-cmfocal lens in place and the laser tuned to resonance with the strong feature at 83 880 cm-I. The maximum flux was estimated to be approximately 1-2 X photons/scm2, and the dependence was measured over a range slightly greater than 1 order of magnitude. The overall trend can be fit reasonably well to a straight line with a slope of unity, as depicted in Figure 8, but there is a clear negative curvature to the data.

Discussion Hydrogen iodide excitation at 239 nm is resonant with the first UV continuum (Figure l), which consists of several excited states correlating with ground-state H atoms and I atoms in either the ground, I(zP,/2), or excited, I*(2P1/2) state. Molecular beam photodissociation s t u d i e ~indicate ~ ~ J ~ that these states are purely repulsive in nature, with a quantum yield for dissociation of unity. Conservation of momentum dictates that virtually all of the released energy is deposited into the kinetic energy of the departing H atom. At 239 nm, the terminal center-of-mass separation velocities for the fragments are -2.0 X 10s A/ns and -1.5 x 10s A/ns for the 1 and I* channels, respectively. For comparison, the ground state of HI has an equilibrium bond length of 1.61 A. Thus, dissociation is rapid on the time scale of our 204s laser pulse in analogy with the directly measured dissociation rate of CH31.1 Two other states not shown in Figure 1 are generally considered unimportant; the 3Z0state is thought to be highly repulsive and lie at elevated energies, and the 3112state cannot be accessed from the ground state in a single-photon process.*O The lowest energy stable electronic states of HI, the b states, begin at 55 833 cm-I (179 nm),21much higher in energy than the range interrogated in the current experiments. Photodissociation at the wavelengths correspondingto the threshold for appearance of the HI+ features yields an I*/I ratio of approximately 1.2, as determined experimentallyIs and theoretically.20 The 128-amufeature we observehas the correct mass signature and appears at wavelengths corresponding to the expected threshold for two-photon production of HI+. We conclude that we are, indeed, generating the parent ion in a 1+1 REMPI processthrough thecontinuum states. The spectral structure detected in the wavelength-resolved spectrum is unlikely to originate in the intermediate states. The first continuum absorption in HI has been carefully considered20*22 and found to be smooth in nature, manifesting no structure. The dependence of the experimental spectrum on the applied electric field indicates that the peaks arise due to relatively long-lived autoionizing Rydberg states that converge to the HI+ ion. The spectral width

Young

13512 The Journal of Physical Chemistry, Vol. 97, No. 51, 1993

B

Figure 9. Kinetic model for the 1+1 REMPI process in HI. The rate constant a describesstimulated absorption/emissionbetween the ground state, X, and the intermediate states, R. The intermediate state can dissociate to products, P, with a rate constant, k ~ or, absorb another photon with a rate, b, to form the ion, C.

of the features is not affected by the stagnation pressure (hence, the temperature of the supersonic expansion), indicating that rotational structure does not play a role. Assuming homogeneous line widths determined solely by lifetime effects, the narrowest peaks autoionize in -3 ps while the strong feature at 83 880 cm-' has a lifetime of -0.2 ps. The HI spectra displayed are obtained with dilute samples under mild expansion conditions. The spectra are invariant with respect to the stagnation pressure (for pressures 1 1 0 psig) and the temporal delay of the pulsed valve. At the lower pressures, alternate ion signals that might be attributed to a cluster reaction are absent. These factors point to the HI monomer as the carrier of the observed signal. Solvent cage effects operational in larger clusters, which might stabilize the dissociating molecule and enhance absorption of an ionizing photon, do not appear to play a role in our results. The appearance threshold of HI+ at the correct IP of the bare molecule implies that any cage effect would involvea purely elastic encounter between the departing fragments and the surrounding cage molecules, an unlikely occurrence. In contrast to our findings, the presence of C H 4 clusters has been postulated to rationalize the appearance of the parent ion in experiments with similar nanosecond laser sources5(vide supra). Our experimental results cannot rule out the contribution of a direct, two-photon excitation mechanism to the observed HI+ signal. However,it is useful to analyze the multiphoton absorption process as two sequential steps with real, as opposed to virtual, intermediate states. The very short lifetime of the intermediate state in the example of HI will, in any case, blur the distinction between direct and sequential excitation. In that light, the ionization rate achieved with our laser source is apparently sufficient to compete with dissociation of HI and yield a measurable parent ion signal. The linear intensity dependence recorded for HI+ indicates that one of the steps is saturated, or nearly so. The initial excitation will not be coherent given the large spectral bandwidth of the continuum absorption feature, which would imply an absence of Rabi population cycling. The same situation likely applies to the ionization step, and furthermore, the HI+peaks reveal no evidenceof laser induced broadening or shifts. The broad-band dye laser source employed in these investigations will, in any case, possess a short coherence time. A lack of coherence in the multiphoton mechanism would indicate that a simpIe rate equation model may be instructive in consideringthe data presented here. The use of traditional kinetic equations to describe multiphoton processes has proven to be widely applicable and highly i n f ~ n n a t i v e We . ~ ~have constructed a three-level model, depicted in Figure 9 (we have used, for the most part, the notation of ref 23), with the ground state, X,the repulsive intermediate states, R,and the final ionized state, C. The state, P, represents the fragment products of HI photodis-

sociation. The stimulated absorption/emission and ionization rates are CY = uF and @ = yF,respectively, with associated cross sections IS and y with a photon flux F. The dissociation rate of the repulsive intermediate states is kD. The set of coupled rate equations

x = -d+ aR

(2)

C=@R P=k$

(4) (5)

can be readily solved by applying the method of Laplace transforms. An insightful limiting case is reached when all of the HI molecules in the laser interaction region sampled by the mass spectrometer are either dissociated or ionized. With a flux of F =1 X photons/s.cm2, an HI absorption cross sectionz2of u = 6.9 X cmZ,and a laser pulse width of TL = 20 ns, the condition F u T L 1~

is satisfied and such saturation is likely in our experiments. The kinetic processes will be complete on the time scale of the laser pulse, and the results are equivalent to assuming an infinitely long pulse of flux, F. Thus, we can apply the final value theorem24 to the Laplace transforms with the result for the ionized state, C

c=L x o 8 + kD

(7)

and likewise for the fragment species, P, we have

The number of ions produced is seen to be some fraction of the initial number of ground-state HI molecules determined by the ratioof theionization rate to the total loss rateof theintermediate state due to irreversibleprocesses. The fraction can be considered as the quantum yield for ion production. Similarly, the product of a quantum yield for dissociation and the ground-state number density yields the number of fragment species produced. These results clearly indicate the competition between dissociation and ionization from the intermediate repulsive states. The dissociation rate is expected to be quite large relative to the ionization rate maintained by the nanosecond dye laser so that kD >> 8. The ion concentration is then given by (9) and the laser flux dependence of HI+ will be linear, the result determined through experimentation. We can arrive at an estimate for the ratio, C/Xo,from our experimentalmeasurements and, hence, make an estimate for the quantum yield of ion formation, @/kD. A typical signal for strong features in the HI+ spectrum is 100-500 mV, prior to amplification. Test data for our microchannel plate detector allow us to calculate a gain of 3.6 X lo7 at the relevant operating conditions. The temporal width of the observed signal is about 50 ns, slightly larger than the laser pulse width, and the MCP is terminated into 50 Q, so that we calculate approximately 20-100 ions collected per laser pulse. A seeded expansionof 0.2% HI in helium at a total backing pressure of 1 psig entails a value of approximately 10" cm-3 for the concentration of HI in the laser interaction region. The effective volume determined by the focused laser beam and the aperture of the TOF is calculated to be 2 X l v c m 3 . Thus, @/kD (20-100)/(2 x 107) = 1-5 x 10-6. Clearly, our ability to

-

1+1 Resonant Multiphoton Ionization

The Journal of Physical Chemistry, Vol. 97, No. 51, 1993 13513

monitor the production of HI+ from the MPI process is a tribute to the sensitivity of our instrument. We can estimate the effective kD from a simple, classical consideration of the dissociation process. The HI bond length, R, will temporally evolve on the excited-state potential, V(R), according to

where a is the center-of-mass acceleration and p is the reduced mass and any initial kinetic energy due to zero-point motion is neglected. The empirically determined potentialsIs are of the form V(R) = A exp[-B(R - R,)]

+C

The constant C corresponds to the asymptotic energy for the I or I* channel, and &is the equilibrium bond length of the ground state of HI. We consider only the I l l state (A = 3.21 eV, B = 2.86 A-I) for the I channel and the 3110state (A = 1.53 eV, B = 4.80 A-I) for the I* channel since these two states dominate the absorption cross section at the threshold wavelengths. The high-lying Rydberg states accessed by two-photon excitation will have potentials very similar to the ion and, hence, very similar vibrational wave functions. In turn, the X 2113/2 state of HI+ has a similar potential to the neutral ground state (Figure l ) , since a nonbonding electron is removed. The HI will be predominantly in the u” = 0 level, even if the vibrational temperature of the expansion is as high as room temperature. Thus, vertical transitions from R = Rewill be the most important. Furthermore, we suppose that dissociation is complete when the system has stretched to the point R = 2R,, since the reduction in the FranckCondon overlap and the increased kinetic energy of the HI will preclude further absorption to autoionizing Rydberg states. Integration of the rate equation with these limits yields a dissociation time, 7 ~ . of

We calculate values of TD = 6 fs for the I l l state and TD = 6.7 fs for the state, or an effective dissociation rate of kD = 1/?D = 1.5 X loi4S-I. Using our estimate for kD and a peak flux of 1026photons/s.cm2, the cross section for ionization originating from the repulsive intermediate states is calculated to be 7 = 1.5-7.5 X 10-18cm2. In contrast, the cross section for singlephoton ionization from the ground state is in the range of 10-75 X 10-18cm2?5 larger than our estimate by 1 order of magnitude. The exact solution to the kinetic equations can be used to model the flux dependence of HI+ using the above estimates for the various constants and letting the laser flux, F,serveas an adjustable parameter. A reasonable approximation of the data, revealing the same curvature and overall linear slope, assumes a maximum flux of =6 X photons/sm2, close toour experimental estimate of 1-2 X photons/s.cm2. A more complicated model which includes the possibility of stimulated emission from the autoionizing states and takes into account the rate of autoionization displays the same basic behavior as the simple kinetic scheme described. An alternative kinetic mechanism can be formulated to explicitly consider contributions from a direct two-photon process, in addition to sequential excitation. The two-photon ionization will then compete with absorption to the intermediate state while the sequential ionization step must contend with dissociation. The rate of dissociation is much larger than the sequential ionization rate, and thus, the solution, in the limiting case described above, indicates that ion formation depends on the relative rates of two-photon ionization and absorption. Since absorption leads to rapid fragmentation, these results are

TABLE 1: Assignment of HI Autoionizing Rydberg Seriee h Huadf Case (e) Limit’ series

assignment

1 2b 3 4 5 6b Assignments taken from ref 28. Notation explained in text. Series not observed in current work.

analogous to the above model as both show a competition between ionization and dissociation of the intermediate state. The identities of the autoionizing Rydberg states that appear in the HI+ spectrum remain to be clarified. Spin-orbit autoionization in HI has been thoroughly investigated and serves as a model forstudiesofautoionizationin molecular systems.26 Severd investigator^^'*^^ have employed nonlinear four-wave mixing techniques to conduct high-resolution (A; 0.5 cm-I) VUV photoionization studies of HI, and several Rydberg series have been identified. Note that our spectral resolution, achieved with a conventional frequency doubled dye laser without an etalon, is comparable to that reported in the four-wave mixing experiments. Most of the features observed in these studies are due to series which converge to the 2 1 1 ~ / 2state of HI+ but autoionize to the 2n3/2continuum due to spin-orbit interaction. In addition, a number of unassigned peaks are ascribed to vibrationally autoionized Rydberg states converging to the u’ = 1 level of the ion core, most notably the ’n3/2 state. The appearance of our spectrum isqualitatively different from theresultsof thesestudies, and we have not attempted a detailed spectral analysis of the current data. However, we have been able to plausibly identify several features as members of the Rydberg series detected in the single-photon work. Using the series labels of ref 28, we have assigned peaks in Figure 3 due to series 1,3-5. Lefebrve-Bri~n~~ has outlined a treatment of the HI Rydberg spectrum in terms of a transition from Hund’s case (c) coupling to Hund’s case (e) as the principal quantum number increases. In Hund’s case (e) coupling, the orbital and spin angular momentum of the energetic Rydberg electron become decoupled from the molecular axis but are still strongly coupled to each other due to the spin-orbit interaction. Mank et a1.28have used the Hund’s case (e) analysis to identify the observed Rydberg series according to the orbital and total angular momentum quantum numbers of the Rydberg electron, 1 andj, respectively, and the rotational level of the ion, J+, to which the series converges. Using the notation of ref 29, J’-J’’AJ(J”),lj, the series are labeled in Table 1. The features we observe are due to transitions to nominally s- and plike Rydberg orbitals. The single-photon studies were able to record well resolved spectral features up to the 2111/2 (v = 0) thresholdI5 of HI+ at 89 122 cm-I, higher in energy than the observed limit of the twophoton spectrum, 4 8 700 cm-’. The overall decrease in the HI+ signal in the two-photon spectrum at shorter wavelengths may be due to the general decrease in the ionization cross section30 concurrent with an increase in the HI dissociation rate as the first step terminates on steeper portions of the repulsive potential. The ion quantum yield, +/kD, decreases, and an insufficient number of ions are produced to be detected even with the very sensitive MCP detector. In addition, resonance requirements for each transition in the two-photon excitation mechanism may result in a contraction of the Franck-Condon accessible region relative to single-photon absorption. Figure 4 shows an expanded view of features due to series 3-5 with principal quantum number n = 6 and is typical of the detail that is observed in our spectrum. The splitting AE between the series 4 and 5 lines should, in the absence of perturbations,

-

Young

13514 The Journal of Physical Chemistry, Vol. 97, No. 51, 1993

correspond to

AE = 4B'- 28'' (13) The ground-state HI rotational constant, E"= 6.34 cm-I,3I leads to a value of E' = 6.42 cm-l for the Rydberg state. A similar rotational constant of 6.4-6.5 cm-1 was found to describe the VUV photoionization spectrum,27,32 and our result also agrees well with a conventional absorption study of HI Rydberg states below the ionization thresholdz1which yielded B'= 6.2-6.4 cm-l. A theoretical study of spin-orbit autoionization in HI26 used multichannel quantum defect theory to calculate the resonance widths of the various Rydberg peaks as a function of the principal quantum number. The 3/2R(0),s (series 4) feature in Figure 4 has a spectral width of 4.5 cm-I for n = 6, which is in excellent concordance with the theoretical value of 5.0 cm-I. Likewise, the 3/2R(O),p (series 1) peak with n = 6 has a measured width of 35 cm-I compared to the calculated value of 33 cm-I. Interestingly, the spectral width of the n = 6, 3/2R(0),speak in our two-photon spectrum (4.5 cm-1) is much narrower than the 9-12 cm-l width observed in the VUV s t u d i e ~despite ~ ~ , ~laser ~ line widths of S0.75 cm-I. Further differences in the fine structure observed in our spectra compared to the single-photon data may be ascribed to the operation of two-photon selection rules. For instance, 0-and S-branches will become allowed. The rotational line strengths will also depend on the two-photon nature of the transition. Noticeably absent from our spectrum are peaks due to series 2 (5/2R(0),dj= 5/z)and series 6 (3/2R(0),dj= 5/2), which are clearly observed in the one-photon studies. In fact, the strong, broad peaks of series 2 are the most prominent spectral features in the one-photon studies. Such different behavior may be a result of the two-photon process operative in our study versus the one-photon transitions recorded in the VUV investigations. Lefebvre-Brionhas expressed the Hund's case (e) wave functions for the missing Rydberg series in terms of case (a) wave functions.29 Consideration of the two-photon selection rules with resonant intermediatestatesof IIor I:character doesnot ruleout transitions to Rydberg states corresponding to these series. The transition probability,P, for a resonant two-photon excitation is proportional to the product of transition dipoles,

investigators using one-photon excitation shows that specific Rydberg series are suppressed in the two-photon spectrum, possibly due to interference effects. A kinetic analysis indicates that the initial excitation step is saturated and that the ionization cross section, due to a free-bound transition, is relatively large. While the dissociation rate of the excited HI is extremely rapid, the flux from a nanosecond laser is still sufficient to ionize a small fraction prior to fragmentation. It should be possible to examine lower lying electronic states of HI using REMPI schemes, similar to the studies reported for HCl,3S even though many of these states are strongly predissociati~e.~~*~~ The reaction chemistry of HI dimers and clusters, observed in the course of our work, presents another interesting avenue of i n v e s t i g a t i ~ n . ~ ~ , ~ ~ Acknowledgment. The author would like to thank Prof. P. Kleiber and Prof. W. C. Stwalley for helpful comments and suggestions. References and Notes (1) Khundkar, L. R.; Zewail, A. H. Chem. Phys. Leu. 1987,142,426. (2) Jiang, Y.; Giorgi-Arnazzi, M. R.; Bernstein,R. B. Chem. Phys. 1986, 106, 171. (3) Danon, J.; Zacharias, H.; Rottke, H.; Welge, K. H. J . Chem. Phys. 1982, 76,2399. (4) Donaldson, D. J.; Vaida, V.; Naaman, R. J . Chem. Phys. 1987,87, 2522. (5) Sapers, S. P.; Vaida, V.; Naaman, R. J . Chem. Phys. 1988,88,3638. (6) Gadanken. A.; Robin, M. B.; Yafet, Y. J . Chem. Phys. 1982, 76, 4798. (7) Brus, L. E.; Bondeybey, V. E. J . Chem. Phys. 1976,65,71. ( 8 ) Strobel, A.; Lochschmidt, A.; Fischer, I.; Niedner-Schatteburg, G.; Bondybey, V. E. J. Chem. Phys. 1993, 99,733. (9) Zhu, Y.-F.; Grant, E. R. J . Phys. Chem. 1993,97,9582. (IO) Szaflarski, D. M.; El-Sayed, M. A. J . Phys. Chem. 1988,92,2234. (11) Rosenstock, H. M.; Draxl, K.; Steiner, B. W.; Herron, J. T. J. Phys. Chem. Ref.Data 1977, 6. (12) Syage, J. A.; Steadman, J. Chem. Phys. Lerr. 1990, 166, 159. (1 3) Codling, K.; Frasinski, L. J. In Atoms in Strong Fields; Nicolaidcs, C. A., et. al., Eds.; Plenum: New York, 1990; Vol. 212, pp 513-528. (14) Codling, K.; Frasinski, L. J.; Hatherly, P.; Barr, J. R. M. J . Phys. B At. Mol. Phys. 1987, 20, 525. (15) Eland, J. H. D.; Berkowitz, J. J . Chem. Phys. 1977, 67, 5034. (16) Duncan, M. A.; Dietz, T. G.; Smalley, R. E. J . Chem. Phys. 1981, 75. 2118. (17) Chewter, L. A.; Sander, M.; Muller-Dethlefs, K.; Schlag, E. W. J . Chem. Phys. 1987,86,4737.

(18) van Veen, G. N . A,; Mohamed, K. A.; Baller, T.; de Vrics, A. E. Chem. Phys. 1983.80, 113. (19) Clear, R. D.; Riley, S. J.; Wilson, K. R. J . Chem. Phys. 1975, 63,

where the sum is over all possible intermediate states R. The above expression indicates that the various excitation pathways, mediated by the multiple intermediate states, are not independent and can interfere. It is possible, then, that in the case of the absent Rydberg states cancellation of phase due to the presence of cross-terms in the transition dipoles leads to a zero transition probability. Similar behavior has been observed in the two-photon ionization of cesium where excitation at energies lying between the6Pand 7Presonancesleads tointerferenceanda sharpdecrease in the measured cross secti0n.3~A destructive interference effect in HI photoionization may be surprising since there are four accessible intermediate states and accidental cancellation would appear to be unlikely. The influence of interference may be indicative of only two states making a major contribution to the transition probability. Further analysis of thedifferencesobserved between the one- and two-photon spectra may prove enlightening. Conclusion

We have excited HI, isolated in a supersonic expansion, with U V light from a tunable nanosecond dye laser. The parent ion, HI+, is observed due to a 1+ 1 REMPI process, resonant with the repulsivestates that comprisethe first U V continuum of hydrogen iodide. The wavelength-resolvedspectrum reveals sharp features which arise due to autoionizing Rydberg levels converging to the HI+ ground state. Comparison with spectra obtained by other

1340. (20) Levy, I.; Shapiro, M. J . Chem. Phys. 1988,89, 2900. (21) Tilford, S. G.; Ginter, M. L.; Bass, A. M. J . Mol. Specrrosc. 1970, 34, 327. (22) Ogilvie, J. F. Trans. Faraday SOC.1971, 67, 2205. (23) Zakheim, D.; Johnson, P. M. Chem. Phys. 1980.46, 263. (24) Oberhettinger, F.; Badii, L. Table of Laplace Transforms; Springer-Verlag: New York, 1973. (25) Carlson, T. A.; Gerard, P.; Krause, M. 0.;Wald, G. V.; Taylor, J. W.; Grimm, F. A. J. Chem. Phys. 1986,84,4755. (26) Lcfebvre-Brion, H.; Giusti-Suzor, A.; Raseev, G. J . Chem. Phys. 1985.83, 1557. (27) Hart, D. J.; Hepburn, J . W. Chem. Phys. 1989, 129, 51. (28) Mank, A,; Drescher, M.; Huth-Fehre, T.; Bowering, N.; Heinzman, U.; Lefebvre-Brion, H. J . Chem. Phys. 1991, 95, 1676. (29) Lefebvre-Brion, H. J. Chem. Phys. 1990, 93, 5898. (30) Carlson, T. A.; Fahlman, A.; Keller, P. R.; Taylor, J. W.: Whitley, T.; Grimm, F. A. J . Chem. Phys. 1984,80, 3521. (31) Huber, K. P.; Herzberg, G. Comtants ofDiatomic Molecules; Van Nostrand Reinhold Co.: New York, 1979; Vol. IV. (32) Hart, D. J.; Hepburn, J. W. In Multiphoton Processes; Smith, S . J., Knight, P. L., Eds.;Cambridge University Press: Cambridge, U.K., 1987; pp 173-183. (33) Huth-Fehre, T.; Mank, A.; Drescher, M.; Bowering. N.; Heinzmann, U. Phys. Rev. Lett. 1990, 64, 396. (34) Morellec, J.; Normand, D.; Mainfray, G.; Manus, C. Phys. Reu. Lett. 1980, 44, 1394. (35) Arepalli, S.; Presser, N.; Robie, D.; Gordon, R. J. Chem. Phys. Lett. 1985, 118, 88. (36) Ginter, M. L.; Tilford, S. G.; Bass, A. M. J . Mol. Specrrosc. 1975, 57, 271. (37) Buntine, M. A,; Baldwin, D. P.; Zare, R. N.; Chandler, D. W. J . Chem. Phys. 1991, 94, 4672. (38) Young, M. A. To be published.