Wavelength dependence of the photochemistry of hydrogen iodide

---

J . Phys. Chem. 1989, 93, 5834-5839

5834

We shall be seeking additional evidence about this mechanism by investigating the branching yield as a function of photolysis ~avelength.~

Chemical Sciences Division, of the U S . Department of Energy under Contract No. DE-AC03-76SF00098. Registry No. HI, 10034-85-2; C2H2, 74-86-2; CH2CHI, 593-66-8; HCCI, 14545-08-5; C2D2, 1070-74-2; DI, 14104-45-1; DCCI, 1454509-6.

Acknowledgment. This work was supported by the Director, Office of Energy Research, Office of Basic Energy Sciences,

Wavelength Dependence of the Photochemistry of Hydrogen Iodide-Acetylene Complexes in Solid Krypton Samuel A. Abrasht and George C. Pimentel*J Laboratory of Chemical Biodynamics, Lawrence Berkeley Laboratory, University of California, Berkeley, California 94720 (Received: January 27, 1989)

Hydrogen iodide-acetylene complexes in a krypton matrix at 12 K have been photolyzed at fixed wavelengths in the range 222-308 nm. Products are identified with infrared spectroscopy. With deuterium substitutions (HIC2D2and DI/HI/C2H2), three product channels are identified leading to isotopic exchange, formation of vinyl iodide, and formation of iodoacetylene. At wavelengths longer than 300 nm, only iodoacetylene is formed whereas at shorter wavelengths, isotopic exchange occurs (e.g., HI.C2D2 DIC2HD), vinyl iodide is formed, and some iodoacetylene is formed. Because the T-shaped geometry of the HI-C,H2 complex is well established, it is possible to interpret these results by recognizing that the photolysis involves two molecules in proximity, a supramolecule. As the hydrogen atom leaves the iodine atom, it forms initially C2H3. Correlating the energy levels along this initial reaction coordinate indicates the initial photolytic reaction surfaces, and Franck-Condon arguments explain the wavelength dependence of the product distribution. This supramolecule analysis provides a prototype analysis for other examples of photolysis of bound and geometrically constrained complexes.

-

Introduction

Photolysis of hydrogen iodide-acetylene complexes in solid krypton results in two products, iodoacetylene and vinyl iodide, as well as exchange in isotopic mixtures.' This outcome raises the question of the extent to which the excitation behavior is a function of the six-atom entity defined by the HIC2H2complex. Even though the complex is sufficiently weakly bound that the ground state can be regarded as two weakly interacting molecular components, it is quite unlikely that this situation persists for the excited species. Instead, the upper electronic states and their chemistries are more likely to be properties of the six-atom system, which we will call the supramolecule. To test this point of view, we have investigated the effect of excitation wavelength on the branching for this hydrogen iodide-acetylene system. Tuned laser photolysis has been used over the wavelength range 222-309 nm; most attention has been focused on isotopic combinations that help identify primary and secondary photolytic processes. Experimental Section

The cryostat, spectrometer, mixture preparation, chemicals, and deposition conditions were identical with those described in paper 1 . I Photolysis conditions differed as described below. The laser was a Quanta Ray YAG pumped dye laser. The pump laser was a DCR-2A optimized for a repetition rate of I O Hz. The 1064-nm beam from the DCR was directed into a harmonic generator (Quanta Ray Model HG-2), where the 532-nm green beam was generated, and through the Quanta Ray prism harmonic separator (PHS-I), where the green beam and the fundamental were spatially separated. The green beam and, in some applications, the 1064-nm beam were then directed through different ports into the dye laser (Quanta Ray, Model PDL). The green beam was then used to pump the dye beam which was directed into the Quanta Ray wavelength extender 'Present address: Department of Chemistry, University of Pennsylvania, Philadelphia, PA 19123. *Deceased June 18, 1989.

0022-3654/89/2093-5834$01.50/0

TABLE I: Spectral Features of Hydrogen Iodide-Acetylene Photolysis Products in Solid Krypton' Y,

CH2CHI

HCCI HCCCCI I-H*-C=C I-D*-C=C

cm-'

1587.2 1386.2 1237.3 952.1 3305.8 2053.5 625.7/624.1/623.8 3314.6 1234.9 625 2179.4 1562.4

u,

cm-'

HCCD

3328.5 2577.7 5 16.4

DCCI

2584.0 1975.2

HI (isolated) DI (isolated)

2231.5 1599.5

(WEX). The WEX was used to double the dye beam and, in some cases, to sum the resulting UV photons with the 1064-nm YAG fundamental. The final beam was then directed through four UV quartz right-angle prisms (Optics for Research and Oriel) to the sample. Quanta Ray specifications for pulse width and pulse to pulse energy variance were 10 ns and 5%, respectively. The specification for the line width was 0.25 cm-'. All laser dyes were obtained from Exciton; they included Rhodamine 590, Sulforhodamine 640, DCM, and LDS 698. Methanol was used as solvent (Mallinckrodt, Spectro Grade or Semiconductor Grade). These dyes provided a wavelength range of 276.2-356.7 nm for the dye second harmonic and from 222.4 to 267.1 nm for the second harmonic + 1064 nm. The wavelength dial on the dye laser was calibrated by using a monochromator (McKee Pederson Instruments, Model MP1018) which was, in turn, calibrated to AO.1 nm by using the second harmonic of the YAG (532.0) and the red beam of a He-Ne laser (632.9 nm). Average power was measured with a power meter (Scientech, Model 380105) optimized for operation between 200 and 400 nm, whose output was read from a digital multimeter (Keithly, Model ( 1 ) Abrash, S . A,; Pimentel, G.C. J . Phys. Chem., preceding paper in this issue. Hereafter, this paper will be called paper I .

0 1989 American Chemical Society

The Journal of Physical Chemistry, Vol. 93, No. 15, 1989 5835

Photochemistry of HI.C2H2 Complexes in Solid Kr

TABLE 11: Photolysis of HI/C2D2/Kr = 1/1/100 at 304.2 and 236.6 nm (Growth Rates in mAU/mmol hv)

photolysis no. 1 2 photolysis A, nm 304.2 236.6 photolysis time, min 60 10 -500

*g -53

6 a?

E -1000

2

584.0, HD DCCI '

-281.2 nm 25

50

Photolysis time (min)

Figure 1, Photolysis rates in experiment A, HI.C2D2: wavelength alternated between 276.2 and 281.2 nm.

177 or 197). Photolysis power was measured approximately 30 cm in front of the sample and outside of the FTIR purge region. Power was measured immediately before and immediately after each photolysis with an uncertainty of 65%. The duration of the photolysis was uncertain to f 4 s. The unfocused laser beam was positioned to maximize the overlap between the laser beam and the IR beam of the FTIR by directing it through two templates to fix the angle of approach and position of the laser beam. Because of differences in the transparency and scattering of different samples, we photolyzed each sample at two wavelengths which were alternated throughout each experiment. Product peak intensities were measured after each photolysis and normalized to the number of incident photons.

Results Table I lists spectral features of the photolysis products, as identified in paper 1 . We shall present experiments here to investigate the wavelength dependence of the four conclusions reached in paper 1, that vinyl iodide is a primary product that undergoes secondary photolysis, acetylenic isotopic exchange is a primary process, and iodoacetyleneis a primary product in which the hydrogen (deuterium) atom from the hydrohalide is preferentially eliminated in the molecular hydrogen product. Ten experiments were conducted in which photolyses were alternated between two different wavelengths. Six of these involved HI/C2D2 mixtures, two DI/C2H2, one HI/C2H2, and one DI/ C2D2. Two of the HI/C2D2and the DI/C2H2experiments will be described in detail. Specific results will be drawn from the others, each of which is described in detail in ref 2. HI.C2D2: Photolysis at 276.2 and 281.2 nm. This sample was photolyzed alternately at 276.2 and 281.2 nm, each at 1 mJ/pulse, or on average, 1.8 X 10l6 photons/s. Figure 1 shows a plot of the loss of 2179.7 cm-' (I-H.-C=C) together with the growths of product absorptions 3327.7 (DI-HCCD) and 2584.0 cm-' (HDeDCCI). The doubled connecting lines denote photolysis at 276.2 nm and the single lines at 28 1.2 nm. A smooth curve through the growth curve data for 2584.0 cm-' displays the same sigmoidal behavior observed in paper 1 (see Figure 4b) for broad-band photolysis and no striking difference in photolysis rates at the two wavelengths. The average rate of production of HDsDCCI for the first eight photolyses is 3.2 X lo4 absorbance units (AU) per minute or 0.068 AU/mmol photons. (2) Abrash, S. A. Ph.D. Dissertation, University of California, Berkeley, 1987.

photons, mmol 2179.7 cm-l IH...C=C 1562.5 cm-l ID. ..C=C 3327.8 cm-l DlaHCCD 2577.5 cm-' DI-HCCD 1534.0 cm-' CHDCDI 3305.8 cm-l Dy HCCI 2583.7 cm-l HD-DCCI

3 304.2 120 0.103 0.0125 0.186 -126 -5100 -28

4 236.6 10 0.0115 -3170

5 304.2 121 0.198 +22

6 236.6 10 0.0142 -2350

+8

+I84

+IO

0.0

+9

-155

+I7

+I030

-22

+960

-34

+I055

+8

+1360

-15

+990

-30

+845

+4

+190

-3

+I60

-6

+50

+6

+I04

+42

+I82

+48

+I35

+9

+I52

+I14

+235

+I11

+I34

In contrast, the loss of 2179.7 cm-I ( I - H - m ) shows distinct alternation, with faster photolysis at the shorter wavelengths. Considering only the first eight photolyses, the four at 276.2 nm average at 0.94 f 0.38 AU/mmol hv while the four at 28 1.2 nm average at 0.41 f 0.21 AU/mmol hv. In a similar way, the growth of 3327.7 cm-I, DI-HCCD, shows a clear difference between the two wavelengths. The first four photolyses at 276.2 nm average at 0.21 f 0.06 AU/mmol hv while those at 281.2 nm average at 0.1 1 f 0.04 AU/mmol hv. In this same experiment, the other prominent absorption of DI-HCCD, 2577.5 cm-', showed behavior quite analogous to that of 3327.7 cm-I, with average growth rates at 276.2 nm of 0.19 AU/mmol hv and, at 281.2 nm, 0.075 AU/mmol hv. Relatively little D2.HCCI was formed (3305.5 cm-I); the average growth for the first eight photolyses was 0.02 AU/mmol hv, with no discernible wavelength dependence. Finally, isotopic exchange was evident in the continual growth of the feature at 1562.2 cm-l due to I-D-.C=C. The growth paralleled that of loss of the I-H.-C=C counterpart at 2179.7 cm-I; the average growth at 276.2 nm (four photolyses) was 0.075 AU/mmol hv and, at 28 1.2 nm, 0.038 AU/mmol hv. If we assume that the extinction coefficient of I-D.-C=C is half that of I-H-C=C, as would be expected for a vibration dominated by the hydrogen atom, then we can estimate the fraction of the lost I-H.-C=C that is converted into I-D-C=C. This average for the first eight photolyses is 20% with no obvious dependence either on photolysis wavelength or on time of photolysis. It is implied that the reincludes formation of maining 80% of the loss of I-H-C=C HCCI, DCCI, and dideuteriovinyl iodide. It implies, furthermore, that we can calculate the effective extinction coefficients of the HCCD features at 3327.7 and 2577.5 cm-I relative to that at 2179.7 cm-' by assuming that 20% of the loss of 2179.7 cm-' formed HCCD. The effective extinction coefficients so derived are t

[ (2 17 9.7) HI*C=C]

t [ (3327.7)DI.HCCDI

= 0.78

6[(2 179.7)HI*C=C] = 0.87 €[(2577.5)DI-HCCD]

HIC2D2: Photolysis at 304.2 and 236.6 nm. This sample was photolyzed alternately at 304.2 and 236.6 nm, each at about 1 mJ/pulse. Average power levels were used to calculate the total number of millimoles of photons incident on the sample during each photolysis period. Table I1 lists the observed spectral changes for seven key features. Both photolysis rate and product distribution are strikingly affected by the photolytic wavelength. For example, Figures 2 and 3 show the spectral changes caused by two successive photolyses, the fifth, at 304.2 nm, and the sixth, at 236.6 nm. Figure 2 shows that the feature at 3305.8 cm-' (due to HCCI) grows markedly during the 304.2-nm photolysis whereas the 3328.4-cm-' absorption (due to DI.C,HD) is lost during

5836 The Journal of Physical Chemistry, Vol. 93, No. 15, 1989

' 7TABLE III: Photolysis of HI/DI/C2H,/Kr = 0.35/0.65/1/100 at hv 46

I

304.2 and 236.6 nm (Growth Rates in mAU/mmol hv)

236 6 nm

1

-0OlOC 3340

7

3330 3320 3310 Wavenumbers (cm-')

3300

Figure 2. Difference spectra in experiment B, HIC2D2. Top: after sixth photolysis, 10 min at 236.6 nm (total photolysis, 301 min at 304.2 nm, 30 min at 236.6 nm). Bottom: after fifth photolysis, 121 min at 304.2 nm (total photolysis, 301 min at 304.2 nm, 20 min at 236.6 nm). I

I

2577 4

oolol K p L

0 000

V

2584.9 I

W

I

II

Abrash and Pimentel

,

I

2600

I

2577 4 1 2580

I

1

2560

Wavenumbers (cm-')

Figure 3. Difference spectra in experiment B, HIC2D2: conditions same as in Figure 2.

304.2-nm photolysis but grows during 236.6-nm photolysis. Figure 3 shows similar contrasts: 2583.7 cm-I (due to DCCI) grows during 304.2-nm photolysis while 2577.4 cm-' (due to DIC,HD) grows during 236.6-nm photolysis. Figure 3 also shows that the 2583.7-cm-' band has a companion feature at 2584.3 cm-'. The photolysis behavior suggests that they are due to the same molecule, DCCI, in two different matrix sites. Such doublet companion features are also noted for 3305.8 cm-' (at 3308.5 cm-I), for 3327.8 cm-' (at 3326.9 cm-I), and for 2577.5 cm-I (at 2576.3 cm-I). Photolysis rate, expressed in A(AU)/mmol of photons incident on the sample for I-H--C=C (2179.7 cm-l) in the first photolysis at 304.2 nm is 0.1 26 AU/mmol whereas in the second photolysis at 236.6 nm,it is much larger, 5.10 AU/mmol. These rates fall off in subsequent photolyses, as parent complexes are depleted. The absorption feature at 1534 cm-I was quite weak in experiment A and could not be monitored accurately. It is assignable

photolysis no.

1

photolysis A, nm photolysis time, min photons, mmol 2177.7 cm-' IH ...C =C 1562.5 cm-' ID...C=C 3327.8 cm-' DIaHCCD 2577.5 cm-' DIaHCCD 1594.1 cm-' CHDCHI 3305.8 cm-' Dy HCCI 2583.7 cm-' HDeDCCI

304.2 120 0.227 -8.8

2 3 4 236.6 304.2 236.6 5 120 10 0.0060 0.215 0.0139 -283 +I5 -863

5 304.2 120 0.203 +I3

0.0164 -354

-7.5

-1765

+2

-440

0

-67

+2.2

+I115

-6

+510

-28

+220

+1.8

+I365

-5

+540

-31

+244

+1.3

+366

-8

+223

-11

+98

+I8

+583

+255

+316

+265

+213

+I

+50

+27

+86

+52

+61

6 236.6 15

as the C=C stretching mode of dideuteriovinyl iodide, CHD= CDI., In this experiment, this feature grows markedly during photolysis at 236.6 nm and, except for the first photolysis, it is lost under long-wavelength photolysis at 304.2 nm. The absorptions of D,-HCCI (3305.8 cm-l) and HDaDCCI (2583.7 cm-l) have quite different behavior from that of CHD=CDI. Though both features grow during short-wavelength photolysis, they grow equally rapidly at 304.2-nm photolysis. The growth of I-D-.C=C (1562.5 cm-I) is inconsistent, possibly because the DI absorption is broadened because of hydrogen bonding and its intensity is strongly dependent on orientation. A better measure of the progress of the isotoic exchange reaction is provided by the two features of HCCD, 3327.8 and 2577.5 cm-I. Both of them grow markedly under short-wavelength, 236.6-nm, photolysis, and after such growth, both are decreased under long-wavelength,304.2-nm, photolysis. At 3327.8 cm-I, the average growth rate of DI.C2HD under 236.6-nm photolysis in the second, fourth, and sixth photolyses is 1.02 AU/mmol hv. At 2577.5 cm-I, the growth averages at 1.1 AU/mmol over the same three photolyses. In the third and fifth photolyses, 3327.8 cm-I decreases by, respectively, 0.022 and 0.035 AU/mmol hv; presumably the loss rate is increasing because of the net accumulation of this species under 236.6-nm photolysis. H I / D i / C z H z / K r= 0.35/0.65/1/100 Photolysis at 304.2 and 236.6 nm. This experiment parallels experiment B except that the initial photolysis begins mainly with DI.C2H2. Hence we expect HCCI to show larger growth than DCCI, as is observed (see Table 111). Vinyl iodide is again formed during 236.6-nm photolysis and lost during 304.2-nm photolysis (except for the first photolysis) except that CH,=CHI (1 587.1 cm-l) and CHD=CHI (1 594.1 cm-I) are formed instead of CHD=CDI (1 534.0 cm-I). Aside from these expected isotopic changes, the major conclusions of experiment B are corroborated. HCCI and DCCI grow during 304.2-nm photolysis as well as during 236.6-nm photolysis. C z H D and vinyl iodide (CH,=CHI and CHD=CHI) grow during 236.6-hv photolysis and are lost during 304.2-nm photolysis. The photolysis rates are in qualitative but not quantitative agreement with those observed in experiment B. In the first, 304.2-nm photolysis, 2179.7 cm-' decreases at the rate of 0.0088 AU/mmol photons while 1562.5 decreases at 0.0075 AU/mmol. Taking into account the lower extinction coefficient of the DI stretching motion, these together are the equivalent of 0.024 AU/mmol at 2179.7 cm-I, a factor of 5 lower than observed in experiment B. The rates of decrease of 2179.7 and 1562.5 cm-l during the second, 236.6-nm photolysis are 0.283 and 1.77 AU/mmol, respectively. Again combining these, they are the equivalent of a 2179.7-cm-I loss rate of 3.8 AU/mmol photons. This is a factor of 1.33 below that observed in experiment B. The growths of the HCCD absorptions at 3327.8 and 2577.5 cm-I during the second, 236.6-nm photolysis are, respectively, 1 . I and 1.4 AU/mmol, to be compared to the experiment B results, 1.02 and 1.1 AU/mmol. During the third, 304.2-nm photolysis,

Photochemistry of HI.C2H2 Complexes in Solid Kr

The Journal of Physical Chemistry, Vol. 93, No. 15, 1989 5831 I

TABLE IV: Reaction Excitation Spectrum: A(2179 cm-') in AU/mmol Photons A. nm

v. cm-I

A(2179.7 cm-ll

222.4 236.6 276.2 28 1.2 286.2 304.2 309.2

44 964 42 265 36 206 35 562 34941 32 873 32 342

7.8 f 1.4 4.5 0.9 0.85 f 0.1 0.48 f 0.1 0.78 0.082 f 0.05 0.057

i

I

1

I

.T

1

*

3327.8 cm-l decreased at 0.006 AU/mmol photons, to be compared with 0.022 AU/mmol photons in experiment B. Reaction Excitation Spectrum. The experiments performed reveal the wavelength dependence of the effectiveness of photolysis. In the H I C 2 D 2and H I C 2 H 2experiments, this is measured by the loss of the characteristic absorption of I-H-C=C (2179.7 cm-I) per incident photon. The number of incident photons during the photolysis period is calculated as the measured photon flux, photons per second, multiplied by the photolysis duration. For experiments in which both H I and D1 are initially present, the loss of I-D-.C=C (1562.5 cm-I) is expressed in terms of equivalent amount of I-H.-C=C by doubling to take account of the isotope effect on the hydrogen iodide extinction coefficient. Table IV compiles the data and Figure 4 shows the reaction excitation spectrum that results, plotted logarithmically. The steep wavelength dependence reflects the product of extinction coefficient times the total quantum yield for reaction. Wavelength Dependence of Product Branching. More interesting than the change of photolysis rate with increasing wavelength is any accompanying change in product distribution. Because of day-to-day changes in the laser beam shape and intensity distribution (its "burn pattern"), absolute photolysis efficiency is difficult to measure. Such difficulties are sidestepped, however, by examining in each experiment relative product growth rates. Table V shows such ratios, each ratio being proportional to the molar ratio with proportionality constant equal to the extinction coefficients of the two species at the characteristic frequencies used for diagnosis. For example, in experiment A, portrayed in Figure 1, the rate of production of DI-HCCD at 2577.5 cm-' is 0.209 AU/mmol hv at 276.2 nm (average of first, third, and fifth photolyses) while the corresponding rate at 2584.0 cm-' due to HDsDCCI is 0.052 AU/mmol hv. The ratio of these two rates, 4.0, is proportional to the relative rates of production of DI-HCCD/HD.DCCI. In the same experiment, the corresponding data at 281.2 nm (average of second, fourth, and sixth photolyses) give 0.078 AU/mmol at 2577.5 cm-I, 0.053 AU/mmol at 2584.0 cm-I, and a ratio of 1.5. Thus, the shorter wavelength favors exchange to produce DISHC C D over reaction to produce HD-DCCI. The same contrast is even more strikingly displayed in experiment B (see Table 11). At 236.6 nm, the rate of production of 2577.5 cm-' is 1.06 AU/mmol and of 2584.0 cm-I, it is 0.17 AU/mmol, giving a relative rate of production equal to 6.1. The behavior at 304.2 nm is even more dramatic, since the 2577.5-cm-' absorbance now decreases by 0.016 AU/mmol while 2584.0 cm-' increases by 0.091 AU/mmol. The last three columns of Table V present the same sort of information, but now product absorbance changes are given relative to loss of parent HI.C2D2 at 2179.7 cm-I. When such a ratio appears as a positive quantity, it is because borh absorbances increased or both absorbances decreased during photolysis. The data show that, as wavelength increases, the exchange product, DI-HCCD, becomes less and less important relative to

Wavelength (nm)

Figure 4. Reaction excitation spectrum: loss of parent absorption at 2179.7 cm-' per mole of photons as a function of wavelength. I

I

I

I

I

-

- 236 6 nm

0

'65

@- _ _ _ _ _ _ _ _ 230

li C

Figure 5. Initial energy surfaces of the hydrogen iodide-acetylene su-

pramolecule. HD-DCCI. At the longest wavelength, 304.2 nm, DIsHCCD is actually photolyzed faster than it is produced. The last three columns show that the product ratio changes as wavelength increases because the fraction of the photolyzed HI.C,D, that gives HD-DCCI increases markedly at the longest wavelengths, while

TABLE V: Wavelength Dependence of Product Yield Ratios A, nm

photolyses averaged

236.6 216.2 28 1.2 304.2

2, 4, 6 1, 3, 5 2, 4, 6 1, 3, 5

HCCDIDCCI

HCCD/HI*C,D,

257712584 6.1 4.0 1.5 -0.18

257712179 -0.30 -0.21 -0.17 +0.56

CHDCDI/HI.C2D, 153412 179 -0.037 -0.047 -0.45 +o. 10

DCCI/HI*C,D, 258412119 -0.049 -0.052 -0.1 1 -3.2

5838 The Journal of Physical Chemistry, Vol. 93, No. 15, 1989

Abrash and Pimentel

at 304.2 nm, both products DI-HCCD and CHDCHI show net losses during photolysis.

Discussion In paper 1, we deduced that when hydrogen iodide-acetylene complexes are photolyzed in solid krypton, there are three primary reaction channels: formation of vinyl iodide, formation of iodoacetylene, and hydrogen exchange. The present data are all consistent with this scheme, and in addition, they show that the branching is strongly wavelength dependent. This dependence can be characterized by the HI.C2D2product preferences at 236.6 and 304.2 nm.

\

\ '\

\ \

\

l , E I

-@j

HD + DCCI

N

~g Energy us H-I Distance: The Initial Reaction. Figure 5 shows, in the dashed curves, the energies of the ground and two excited states of gaseous HI, 1(I) and O+(II), as calculated by Chapman and c o - ~ o r k e r s . ~We must be concerned with how these curves are modified by the proximity of an acetylene molecule in the geometry of the HI-C2Dzcomplex. The distance between the iodine atom and the center of the triple bond can be estimated by using the systematics compiled by Pimentel and M ~ C l e l l a n . ~ They found that in an A-H--B hydrogen bond the A-B distance is close to or somewhat smaller than the sum of the van der Waals radii of A and B. (assigning no "size" to the H atom). The van der Waals radius of an iodine atom is 2.2 8,, and the extent of the triple bond can be estimated to be half the distance between the planes of graphite, 3.50/2 = 1.75 8,. We will take the sum of these 2.2 1.75 = 3.95 h; to be the length of this hydrogen bond. Another general characteristic of hydrogen bonds is that the A-H separation is increased relative to the isolated A-H molecule.4b For relatively weak hydrogen bonds, this will be a few hundredths of an angstrom, so the H I bond length will be lengthened to about 1.65 8,. With a total iodine atom-acetylene distance of 3.95 A, this would place the hydrogen atom 2.30 8, from the center of the triple bond. We assume, finally, that the hydrogen bond energy in the HI-C2D2complex is about 3 kcal/mol. We begin our analysis with the presumption that initial excitation is to a reaction surface strongly influenced by the dissociative character of the excited states of the parent H I molecule. The implication is that the first dynamical response to excitation will be movement of the HI hydrogen atom away from the iodine atom and toward the acetylene. Because of momentum conservation, the iodine atom will be essentially stationary. Initially, the coupling between the atom and the acetylene will be weak so, at first, the acetylene molecule also will be essentially immobile. Then as the hydrogen atom moves, it has the opportunity to form the stable molecule C2HD2. Thus we see that the general characteristic of the initial reaction surface can be deduced from the correlation diagram that connects the energy surface of isolated H I with that of isolated C2HD2. This correlation diagram depicts the energy surfaces associated with the six-atom "supramolecule", which initially has the structure HI.C2D2. The heat of formation of , ~ the ~ ~ ground state C2HD2is known to be 66 & 6 k c a l / m ~ l so of l(2P3i2)+ C2HD2 lies 30 kcal/mol above H I + C2D2. An electronically excited state of CzH3* has been detected spectroscopically by Hunzinger et al.,' who found a progression that peaks

+

~

~~~~

(3) Chapman, D. A,; Balasubramanian, K.; Lin, S. H. Chem. Phys. Lett. 1985. 118, 192; J . Chem. Phys. 1987, 87, 5325. (4) (a) Pimentel, G. C.; McClellan, A. L. The Hydrogen Bond, Freeman: San Francisco, 1960. (b) Pimentel, G . C.; McClellan, A. L. Annu. Reu. Phys. Chem. 1971, 22, 347. ( 5 ) Lossing, F. P. Can. J . Chem. 1971, 49, 357. (6) DeFrees. D. J.; McIver, Jr., R. T.; Hehre, W. J. J . A m . Chem. SOC. 1980, 102, 3334.

-43I= CDI

CHD

Figure 6. Continued chemistry in the photolysis of HIC2D2: wavelength

dependence. at 24815 cm-' and has an onset near 20020 cm-' (57.2 kcal/mol). In reasonable agreement, Paddon-Row and Pople* have reported ab initio calculations that place the lowest excited state of C2H3 near 18 590 cm-I. Figure 5 shows how these states must correlate to those of HI. They provide a basis for sketching in the likely reaction surfaces associated with the "supramolecule" leading, at first, to I C z H D 2 . The ground state of HI, O+(I), dissociates to form 1(213,2), so this state must correlate with the I(2P3/z)+ CzHD2(X) ground states. Our correlation diagram shows that this occurs on a surface that probably includes an activation energy barrier 10-20 kcal/mol above the endothermicity. The second state, I(2P,j2) C2HD2(X)undoubtedly correlates with the H I O+(II) state, because O+(II) in the gas phase dissociates to I(2P1/2). That in turn leads us to connect the H I 1(I) state to the I(2P31z) C2HD2*state, placing an energy minimum at an I-H distance of about 2.8 8, and perhaps 10 kcal/mol deep. Also shown in Figure 5 are the excitation energies associated with two of the wavelengths we have used, 236.2 and 304.2 nm. As drawn, the energy surfaces imply Franck-Condon bias favoring excitation to the 1(I) state relative to the O+(II) state because of the mismatch between the I-H distances in the ground state, O+(I), and the two excited states. This mismatch is relatively small for 236.6 nm (0.02 and 0.17 8,. respectively), but for 304.2 nm, it is 0.27 8, for l(I) and 0.59 8, for O+(II). This mismatch leads to the expectation that the fraction of excitation to 2P1/zstate should decrease as wavelength increases. Clear et aL9 have summarized the evidence about this fraction as a function of wavelength. Despite large uncertainties, the existing data suggest that the percent I(2P,,2) increases to a maximum of perhaps 50% at 253.7 nm and then decreases at longer wavelengths. This would be in accord with Figure 5, which suggests that for 236.6-nm photons there will be little Franck-Condon bias; hence there can be substantial excitation to the energy surface leading to I(zPl/2) + C,HD2. However, at longer wavelengths, 304.2 nm,the large Franck-Condon mismatch will give a definite preference for excitation to the I(2P312)+ C2HD2*surface. Such a change would lead to a change in product distribution, as observed. Continued Cage Chemistry. While the reaction surface proposed in Figure 5 may usefully depict the initial effect of excitation of H I in the geometry of the HI.C2D2 complex, subsequent chemistry is to be expected. The outcome of the excitation, an iodine atom held in the same cage as the free radical C2HDz,will surely lead to additional reaction. Figure 6 adds two such reaction

+

+

( 7 ) Hunziker, H. E.; Kneppe, H.; McLean, A. D.; Siegbohn, P.; Wendt, H. R. Can. J . Chem. 1983, 61, 993. (8) Paddon-Row, M. N.; Pople, J. A. J . Phys. Chem. 1985, 89, 2768. (9) Clear, R. D.; Riley, S. J.; Wilson, K . R. J . Chem. Phys. 1975, 63, 1340.

Photochemistry of HI.C2H2 Complexes in Solid K r channels, to form HD-DCCI and CHD=CDI. Because of Franck-Condon bias, we conclude that excitation at wavelengths longer than 300 nm excites I(2P3,2) C2HD2* in preference to 1(2Plj2)+ C2HD2. If so, then this excitation apparently leads only to HD-DCCI and we suggest that this might proceed via molecular hydrogen elimination from C2HD2*to give C2D. This molecule then can react with I(2P3j2)in the same cage to give the observed product, HD-DCCI. At wavelengths as short as 236 nm, the smaller Franck-Condon bias permits excitation to the reaction surface correlated with 0+(11), which leads to the I(2P112) C2HD2state. Again experiment shows that this reaction channel leads to the formation of vinyl iodide or return to starting product after isotopic exchange. As shown in Figure 6, vinyl iodide may well be formed through its triplet state with a modest energy barrier easily overcome by the nascent vibrational energy lodged in C2MD2. Finally, Figure 6 suggests how secondary photolysis of vinyl iodide at a wavelength of 304.2 nm would invest enough internal energy to 3[CHD=CDI] to permit formation of I(2Plj2)+ C2HD2 and, then, return to parent with isotopic exchange. Mechanistic Summary. Figure 5 implicates two states of C2H3, the ground, 2A’ state which Paddon-Row and Popleg characterize as a “u radical”, and the first excited 2A’’ state, a “ x radical”. These characterizations are supported by the calculated C-C bond lengths, 1.328 8, for 2A’ and 1.441 8, for 2A’r. It is reasonable to expect their chemistries to be dissimilar. Consider, first, excitation at 236.6 nm where the FranckCondon effects are small so that both 1(I) and O+(II) states are excited. The latter correlates with the 1(2Pl/2) C2HD2(2A’), and it provides a large amount of excess energy, 67 kcal/mol, with which to accomplish isotopic scrambling. If the “ u radical” C2H3 reacts with I(2Pl/2), it can either cool to give vinyl iodide, CHD=CDI, or eliminate hydrogen iodide to give the isotopic mixture, HI-C2D2and DImHCCD. To complete our explanation, it is sufficient to postulate that excitation of the 1(I) state of H I gives the 2A” state of vinyl radical, the “x-radical” configuration, CHDCD*, and that this radical reacts in the cage to give exclusively HD-DCCI. At longer wavelengths, the Franck-Condon effect causes this to be the only reaction channel excited, so the long-wavelength product is exclusively HD-DCCI. While this is a pleasingly economical model, it must be noted that the final reaction must retain the identity of the projectile hydrogen, since only H D is formed. Two reaction channels suggest themselves, each with difficulties. The first possibility is that C2HD2*can lose H D through aa elimination at the C H D end of the molecule where the projectile hydrogen has lodged. After elimination, the C2D radical would combine at leisure with its cage partner to give DCCI. This channel requires that the activation energy for aa elimination be quite low on the excited-state surface. Unfortunately, this explanation is clouded by the uncertainty range presently associated with the heat of formation of C2D since it leaves doubt that there is enough energy for this channel to be open. The second possible channel would be reaction between I(2P312)and C2HD2*to form the first excited singlet state of vinyl iodide. This molecule could, in turn, aa eliminate H D with almost zero activation energy. Again, this explanation suffers from the fact that the available data place ‘[CHDCDI]* just above the available energy. In view of these ambiguities, Figure 6 merely displays the phenomenological outcome: I(2P312)+ C2HD2* produces HDsDCCI.

+

+

+

The Journal of Physical Chemistry, Vol. 93, No. IS, 1989 5839 Comparison to Other Work

It is plain that the concepts embodied in Figure 5 are applicable to molecular complexes formed in supersonic jet nozzles as well as in matrices. However, in absence of the matrix cage, change can be expected in the “continued chemistry” associated with ongoing reaction, as displayed in Figure 6. Furthermore, the two types of experiment provide access to different types of information, so there will be interesting complementarity between matrix studies, like those presented here, and analogous molecular beam studies. In fact, Wittig and co-workers’+l2 have recently published such work, though their interpretation is couched in different language. They photolyzed the complexes DBr.C02 and DBPOCS and examined the energy distributions in the OD and S D products through laser-induced fluorescence. They find distinct changes in these energy distributions in comparison to “bulk reactions”, hydrogen atoms with C 0 2 , and attribute these changes to the “precursor geometry limited” characteristics of the complex. They have not yet investigated wavelength dependencies, and for the OCS case in which product branching occurs, they were not able to detect the S D channel under nozzle conditions. They interpret their results with the premise that “the long van der Waals bond will tend to isolate the excitation to either the O C S or HBr moities”,12but they also express caution about this premise. So while Wittig and co-workers are not overtly recognizing the “supramolecule”, it is plain that their analysis could readily be so extended. We are encouraged to anticipate fruitful and complementary interaction between molecular beam and matrix studies of the special photochemistry of molecular geometries with predetermined geometry. Con c1usions

These experiments show a distinct wavelength dependence of product branching: at longer wavelengths, only iodoacetylene is formed, whereas at shorter wavelengths, isotopic exchange is observed along with formation of vinyl iodide. Clearly vinyl iodide undergoes secondary photolysis, but to give only the parent or isotopically exchanged complex. This wavelength dependence of product branching is clarified by recognizing that the photolysis involves two molecules in proximity, a supramolecule. The photolytic reaction surfaces of this supramolecule can be deduced by correlating the energy levels along the initial reaction coordinate. The Franck-Condon effect then explains why the reaction surface accessed changes with excitation wavelength. We believe that this analysis provides a prototype that will be useful and applicable in other photolytically induced reactions that begin with a bound complex, a supramolecule, whose photochemistry is determined by the proximity and geometry of the two molecular constituents.

Acknowledgment. This work was supported by the Director, Office of Energy Research, Office of Basic Energy Sciences, Chemical Sciences Division, of the U S . Department of Energy under Contract No. DE-AC03-76SF00098. Registry No. HI, 10034-85-2; C2H2,74-86-2; C2D2, 1070-74-2; C2H3, 2669-89-8; vinyl iodide, 593-66-8; iodoacetylene, 14545-08-5. (10) Radhakrishnan, G.; Buelow, S.; Wittig, C. J . Chem. Phys. 1986, 84, 121.

(1 1) Buelow, S.; Radhakrishnan, G.; Wittig, C. J . Phys. Chem. 1987, 91, 5409. (12) Hausler, D.; Rice, J.; Wittig, C. J . Phys. Chem. 1987, 91, 5413.