Structures of bromoalkanes' photodissociation in solution by means of ultrafast extended x-ray absorption fine-structure spectroscopy

Abstract

The structures of initial and final products of bromoalkanes' photodisociation reaction in cyclohexane solution have been measured with a bond length accuracy of 0.02 Å by means of ultrafast time-resolved extended x-ray absorption fine structure spectroscopy. The photoredaction mechanism is also discussed.

The knowledge of structural changes during ultrafast physical, chemical, and biological processes has always been a key factor in understanding the mechanism of most molecular processes. Several techniques, based on the generation of ultrashort electron and x-ray pulses, have been developed during the past decade that provide direct time-resolved structural information of short-lived intermediate species and excited states involved in such processes (1). Most of these techniques use an ultrashort laser pulse to initiate the photoreaction in the sample studied followed by a delayed ultrashort electron or x-ray pulse that probes the changes that take place in the sample. In the solid state, laser-induced heat and strain propagation in metal and semiconductor single and mosaic crystals has been studied by means of nanosecond and picosecond time-resolved x-ray diffraction (2–8). Reflective low and high energy electron diffraction techniques have also been used to study laser-induced surface melting and crystal lattice thermal relaxation with nanosecond and picosecond time resolution (9–12). Excited-state structures of photoactive proteins have been measured by nanosecond time-resolved x-ray crystallography (13, 14), and picosecond x-ray powder diffraction has been used to measure short-lived electronically excited states in organic solids (15). Photoexcited states of molecular iodine in solution were probed by picosecond diffuse x-ray scattering (16). A few gas-phase ultrafast electron diffraction experiments have also been reported (17–20). All of the above techniques provide excellent direct time-resolved structural information; however, usually high order of periodicity of the sample is required. This requirement significantly limits their application in highly disordered systems, such as liquids and amorphous solids. In addition, the diffraction signal has contributions from all molecules in the sample, therefore only neat liquids or very concentrated solutions are suitable for such experiments. There is, however, a technique, extended x-ray absorption fine structure (EXAFS) spectroscopy (21), that may be used for relatively dilute solutions to determine, with high accuracy, the structure of a molecule and its nearby neighbors in solution. EXAFS spectroscopy is based on the analysis of low-amplitude oscillations in the region of 40–1,000 eV higher than the absorption edge of a specific atom in the x-ray absorption spectrum of the material. The structure of the nearest few coordination shells surrounding this atom can be extracted directly. This information, which is very accurate, includes the average distances between the atoms, the average square deviations from these distances (which result from the thermal oscillations and disorder of the system), the coordination number, and type of atoms in each of the shells. In some cases even the high-order anharmonic force cumulants may be measured with acceptable accuracy.

EXAFS has been extensively applied in the study of liquid, solid, and gas structures; however, in almost all of the cases the studies were restricted to ground-state molecules. Several time-resolved experiments were reported with microsecond resolution (22–24). Time-resolved soft (90–300 eV) x-ray absorption spectra have been recorded with nanosecond resolution from the silicon L edges (25) and with picosecond resolution from the Kr M edges (26). Picosecond changes in near edge absorption structure of the sulfur K edge (2.48 keV) in SF₆ gas have also been observed (27). To the best of our knowledge, however, the only ultrafast experiment on hard x-ray EXAFS of liquids have been reported by us (8, 28, 29) and Chen et al. (30). In ref. 30 the excited-state structure of a charge transfer Ni complex has been measured near the Ni K edge (8.34 keV). These very interesting measurements have been performed for one delay time only (picosecond laser pulse coincided with nanosecond x-ray pulse), and no structure kinetics have been measured so far by these investigators. In our earlier papers (8, 28, 29) the bromine K edge (13.5 keV) EXAFS spectra of initial and final products of carbon tetrabromide (CBr₄) photodissociation in ethanol solution have been measured by our nanosecond ultrafast EXAFS system. In this article we extend these experiments to the photodissociation of CBr₄ and several other bromoalkanes, e.g., bromoform (CHBr₃) and dibromomethane (CH₂Br₂), dissolved in cyclohexane. We have measured the EXAFS spectra of the initial and final products of the photoinduced dissociation and extracted from the experimental data their molecular and solvated structures.

The data presented in this article and in refs. 8, 28, and 29 demonstrate that our ultrafast EXAFS experimental system is capable of detecting and measuring very accurately ultrafast changes in the structure of molecules in solution.

Experimental Procedures

The ultrafast time-resolved EXAFS system that we used for the studies considered in this article has been described (8, 28, 29), therefore only a brief description of this system will be presented here. Fig. 1 shows a schematic representation of the ultrafast time-resolved EXAFS system, which uses a nanosecond ArF laser system and a laser driven x-ray diode. For these EXAFS experiments the copper anode was replaced by tungsten, because tungsten is a more efficient x-ray continuum generator. The system is operated in two modes: (i) nanosecond mode, where the ArF excimer laser is used to generate 193-nm, 12-ns pulses and (ii) picosecond mode where the ArF excimer laser is used as an amplifier seeded with 193-nm picosecond pulses (8). Nanosecond EXAFS experiments are easier to perform because the x-ray flux per nanosecond pulse is much higher than within a picosecond pulse. The 193-nm laser pulse was used to produce both a Raman-shifted UV pump pulse for sample excitation and electron pulse that used to generate x-rays. The x-ray continuum generated by the 193-nm laser-pumped x-ray diode has the same energy spectrum as that of a conventional tungsten x-ray tube but a pulse duration of 12 ns at a repetition rate of 300 Hz (8). The divergent x-ray continuum beam was formed by a 50-μm vertical slit situated 60 mm away from the anode. After passing through a 2-mm cell containing the liquid sample, the x-ray pulse was reflected by a 20-cm diameter Si(111) crystal oriented for (422) reflection. The center of the crystal was situated 20 cm away from the slit. An x-ray charge-coupled device detector (1,242 × 1,152 pixels, 22.5-μm pixel size) was positioned 40 cm away from the crystal and was capable of recording, simultaneously, 500 eV or more of the x-ray spectrum. In the course of the experiment the sample was set in a position so that half of the x-ray beam passed through the sample whereas the other half propagated through air only. The x-ray signal that passes through the air is used as reference. This arrangement makes possible the simultaneous recording of both the sample and reference EXAFS signals. The energy resolution of this system was found to be 5 eV, determined by the width of the x-ray source in the vicinity of Br K edge. By selecting a more asymmetric crystal reflection and decreasing the width of the slit an even higher energy resolution is possible.

Figure 1.

Schematic diagram of the ultrafast time-resolved EXAFS experimental system. CCD, charge-coupled device.

The calibration of our experimental system was performed by using the bromine K-edge EXAFS spectrum of 0.4 M aqueous solution of ZnBr₂. The parameters of the first coordination shell of Br⁻ ion extracted from the experimental data were found to be in agreement with the published data (see ref. 29 for details).

A standard automated data reduction procedure (code AUTOBK, see ref. 31) was used to extract the pure experimental EXAFS spectrum. The EXAFS spectrum was also calculated theoretically by using the FEFF8 code (32) and fitted to the experimental one by FEFFIT code (33).

For our experiments, 99% purity CBr₄, CHBr₃, and CH₂Br₂ were dissolved in HPLC-grade cyclohexane to concentrations of 0.15 M, 0.17 M, and 0.31 M, respectively. All sample molecules and cyclohexane were obtained from Aldrich. For each sample we first measured its EXAFS spectrum with only nanosecond x-ray probe pulses. Then each sample was irradiated for 20–23 h with a high-pressure Hg UV lamp, and subsequently we measured the EXAFS spectra of the UV-irradiated samples, keeping all experimental parameters the same. The data collection time for each experiment was between 20 and 30 h. The data presented show the structures of the initial and final products evolved in the bromoalkanes' photodissociation process.

Before time-resolved EXAFS experiments are routinely used, several aspects of this method need to be addressed. For acceptable-quality EXAFS spectra with our experimental system, we need to collect data for 20–30 h. For kinetics measurements the experiment has to be repeated for several delays between excitation and probe pulses, which increases the total time of the experiment to days. Therefore, the x-ray photon flux impinging on the sample must be increased several-fold. One of the possible ways to improve the x-ray flux is to use x-ray polycapillary focusing optics. Preliminary tests of the x-ray focusing system provided by Xunliang Ding from Beijing Normal University show that the x-ray flux on the sample can be increased by 2 orders of magnitude. The use of polycapillary optics has the advantage of removing the higher harmonics, thus allowing the x-ray diode to operate at higher voltages that increase the generated x-ray flux. The second important aspect in time-resolved EXAFS experiments is the proper excitation of the sample. Time-resolved experiments are based on the assumption that the reactions under study are initiated uniformly and simultaneously in the probed volume. This can be achieved only up to a certain degree. The most suitable systems are those that are initiated by optical pulses, because a wide variety of short laser pulses are available. Photoreversible reactions are especially attractive because they allow a reaction sequence to be repeated many times and consequently x-ray scattering data to be accumulated from many shots by using the same sample. At high repetition rates, however, it may require a sample replacement if the reversible process is slower than the repetition rate or there is no full recovery of the system between pulses. Additionally, to achieve the best match between the excitation volume and the penetration depth of the probing x-ray beam, the sample concentration, thickness, and solvent choice must be carefully considered. In the EXAFS studies, the absorption change of the sample at the edge position for the probing x-ray radiation is selected to be in the range of 1 < μx < 3 (34), where μ is the sample's mass absorption coefficient and x is the length of the sample. Using the sample concentrations reported here the corresponding sample thickness was about 2 mm. This thickness has to be uniformly excited by the pumping laser pulse. The 193-nm pulses generated by our laser system are very efficient for x-ray diode pumping but they are not suitable for the excitation of the liquid samples because most solvents absorb at this wavelength. The light excitation pulses used were generated by the nonlinear conversion of a part of the 193-nm radiation to the Raman scattering of the fourth and fifth Stokes lines in hydrogen at 284 nm and 322 nm. This process is quite efficient even at high repetition rates (35). By this means our experimental system can provide both pumping UV pulses suitable for the sample excitation and probing x-ray pulses. The UV and x-ray pulses are very accurately synchronized with each other. Our experimental system is constructed, usually, in such a manner that the x-ray beam is probing a sample volume of 1 × 2 × 0.05 mm³, which is the same as the volume that is exited by the pumping UV laser pulse. For example, if 50% of the CBr₄ molecules must be dissociated (assuming quantum efficiency of 1) 10 mJ at 322 nm radiation must be delivered per pulse, to achieve acceptable uniformity of excitation for a 50% transmitting samples. Experimentally this is possible but rather challenging. In many reactions the processes triggered by the laser excitation pulse are not completely reversible; therefore replacement of the sample after every shot is necessary. For liquid samples the simplest approach that we have used is the flow of the liquid sample. At a repetition rate of 300 Hz a sample flow speed of 0.5–1 m/s is sufficient to perform the experiment.

Results and Discussion

The photodissociation of bromoalkane molecules has been extensively studied both in solution and gas form by techniques such as time of flight MS (36, 37), time-resolved fluorescence spectroscopy (38), pulse radiolysis (39–41), and others. The intermediate products of these reactions have been observed; their structures, however, still remain obscure. In this article we are taking a step in the measurement and elucidation of these structures in solution: we measured the structures of initial and final products of several bromoalkanes' photodissociation in cyclohexane solution.

Fig. 2 shows the Fourier-transformed EXAFS spectra of irradiated and nonirradiated CBr₄, CHBr₃, and CH₂Br₂ in cyclohexane. All three spectra of the nonirradiated molecules show three distinctive coordination shell maxima. Each of these maxima has been filtered and transformed into k-space with subsequent fitting to the theoretical EXAFS data calculated for a molecular structure for each sample. In all calculations we used the 2.2- to 8.7 Å⁻¹ k-range for fitting. The results of the fitting, along with filtering window parameters for each of the shells are listed in Table 1. As a representative example, Fig. 3 shows the CBr₄ three-shell fitting. First, we performed fitting of the first and the third shells, which represent Br-C and Br-Br single intramolecular scattering. For each scattering path the used fitting parameters were the average atomic distance r, its square fluctuation (or Debye–Waller factor) σ², the amplitude factor S, and the shift of energy origin ΔE₀. The coordination number n was set to 1 for Br-C scattering and 3 for Br-Br scattering. The imaginary energy shift and high-order anharmonic cumulants were set to 0. The best fit was obtained with ΔE₀ = 0.8 eV (for both Br-C and Br-Br scattering paths) and amplitude factors S = 0.94 (for Br-C) and S = 1.35 (for Br-Br). These values were then used in all other calculations, for both nonirradiated and irradiated samples. The average atomic distances obtained from the fitting (r_C-Br = 1.91 ± 0.02 Å and r_Br-Br = 3.13 ± 0.01 Å) are slightly shorter than the distances obtained by electron diffraction and gas EXAFS (42) experiments, but the deviation is within the experimental error. The Debye–Waller factors σ² were found to be 0.003 ± 0.003 Ų and 0.016 ± 0.004 Ų for Br-C and Br-Br distances, respectively.

Figure 2.

Radial distribution function of CBr₄ (a), CHBr₃ (b), and CH₂Br₂ (c) before and after UV irradiation. FT, Fourier transformed.

Table 1.

Results of three-shell fitting of irradiated and nonirradiated bromoalkanes

Molecules	Data	First shell (C-Br)		Second shell (Br-Br)		Third shell (Br-Br)
		Not irrad.	Irrad.	Not irrad.	Irrad.	Not irrad.	Irrad.
CBr4	R	1.91  ± 0.02	1.89  ± 0.02	3.07  ± 0.01	3.05  ± 0.01	3.13  ± 0.01	3.19  ± 0.01
	σ2	0.003  ± 0.003	0.006  ± 0.003	0.008  ± 0.001	0.011  ± 0.001	0.016  ± 0.004	0.009  ± 0.001
	n	1	1	2.7  ± 0.2	4.7  ± 0.3	3	1.4  ± 0.1
	r-range	1.11–1.80	1.11–1.70	1.80–2.40	1.70–2.30	2.40–3.40	2.30–3.50
CHBr3	R	1.90  ± 0.02	1.87  ± 0.02	3.06  ± 0.01	3.06  ± 0.01	3.12  ± 0.01	Bad fit
	σ2	0.003  ± 0.004	0.005  ± 0.002	0.009  ± 0.001	0.009  ± 0.003	0.014  ± 0.01
	n	1	1	3.3  ± 0.2	4.0  ± 0.5	2
	r-range	1.11–1.80	1.11–1.70	1.80–2.40	1.70–2.35	2.40–3.45
CH2Br2	R	1.90  ± 0.02	1.91  ± 0.03	3.07  ± 0.01	2.96  ± 0.05
	σ2	0.002  ± 0.003	0.005  ± 0.008	0.013  ± 0.004	0.000  ± 0.001
	n	1	1	5.4  ± 0.4	0.7  ± 0.8
	r-range	1.11–1.70	1.11–2.40	1.70–2.40	1.11–2.40

R, average distance (Å); σ², Debye–Waller factor (Ų); n, coordination number; r-range, filtering window range (Å). 

Figure 3.

Br K-edge EXAFS spectrum of nonirradiated CBr₄/cyclohexane solution after first-shell (a), second-shell (b), and third-shell (c) filtering. Points are experimental results, solid lines are theoretical calculations.

The second shell most likely originates from Br-Br intermolecular scattering. It was shown by neutron scattering that in neat liquid form CBr₄ molecules associate into complexes with 13 ± 1 molecules in the first coordination layer (43). In solution, we expect to see similar structure with a lower number of molecules in the first coordination layer. To prove this notion we used all of the parameters of the third-shell Br-Br single scattering fitting, varying only the average distance r, Debye–Waller factor σ², and the coordination number n. As seen in Fig. 3b the Br-Br scattering calculations fit perfectly well with the experimental data. The average intermolecular Br-Br distance obtained from this fit is equal to r_Br-Br = 3.07 ± 0.01 Å and is consistent with the neat liquid neutron scattering data (43). The coordination number was found to be equal to 2.7 ± 0.2. Because each of the intermolecular bromines in the second shell is most likely shared between at least two intramolecular bromines, and not all of the intermolecular bromines in the coordination shell are from different CBr₄ molecules, the number of CBr₄ molecules in the first coordination layer is definitely much less than 13, the neat liquid value. We performed similar fitting procedures for CHBr₃ and CH₂Br₂ EXAFS spectra. Before the fitting, the spectra were normalized to have the same first-shell peak intensity as in nonirradiated CBr₄. We believe that necessity of the normalization is due to the fluctuations in the total number of x-ray photons collected in each of the measurements. These fluctuations directly affect the intensity of EXAFS oscillations. In the theoretical model used for the fitting, the EXAFS intensity is proportional to the amplitude factor S; therefore, by adjusting this parameter we can account for the x-ray photon number fluctuations. However, since the first shell, in all nonirradiated samples, originates from a single Br-C scattering with approximately the same average distance and Debye–Waller factor, it is prudent to normalize them to yield the same S for the first shell to compare the EXAFS spectra from different samples. The results of fitting are presented in Table 1. The single Br-C (first shell) and Br-Br (second and third shells) scattering models fit the experimental data quite well, except for the third shell in the CH₂Br₂ molecule. Even though a relatively good fit can be obtained by addition of one extra single Br-Br or Br-C scattering path, it is difficult to realize its physical meaning. Therefore, at the present time without performing more experiments, we will not discuss these results. The average distances for all shells in the nonirradiated molecules obtained by the data fit were found to be the same, within experimental error. The coordination number n of the second shell, which represents intermolecular bromine scattering, is increasing with a decreasing number of Br atoms per molecule: n = 2.7 ± 0.2 in CBr₄, n = 3.3 ± 0.2 in CHBr₃, and n = 5.4 ± 0.2 in CH₂Br₂.

The EXAFS spectra of the irradiated samples have been measured by exactly the same experimental procedure as the nonirradiated samples. As in the case of the nonirradiated samples, all EXAFS spectra were normalized to have the same first-shell peak intensity as in the nonirradiated CBr₄. Since the shape and the position of the first shell in all samples do not change much after irradiation (see Fig. 2), we believe that this shell may still represent single Br-C scattering. In addition, a bromine atom can be bounded with only one carbon atom, therefore the coordination number is always equal to 1 for the first shell. As seen in Fig. 2 irradiated samples still have a three-shell peak spectrum as the one we found in nonirradiated samples. The positions of the peaks and their intensities, however, are different. We performed fitting with the same single scattering models as for the nonirradiated samples, with the only difference that in the third-shell fitting the coordination number n was not fixed. As a representative example, Fig. 4 shows the three-shell fitting in irradiated CBr₄. In an irradiated CH₂Br₂ molecule, the second-shell peak appears as a shoulder in the right side of the first-shell peak. Because of this we did not separate the first and the second shells in the fitting for this molecule, but treated them as a single shell, with two single scattering paths (intramolecular Br-C and intermolecular Br-Br scattering). The results of this fitting can be found in Table 1.

Figure 4.

Br K-edge EXAFS spectrum of UV-irradiated CBr₄/cyclohexane solution after first-shell (a), second-shell (b), and third-shell (c) filtering. Points are experimental results, solid lines are theoretical calculations.

The initial process induced in the bromoalkane samples by UV irradiation is most likely the photodissociation of Br atoms. Indeed, the intensity of the third-shell peak representing intramolecular Br-Br scattering strongly decreases after irradiation of CBr₄ and CHBr₃; consequently the shell coordination number decreases (see Fig. 2). The decrease in intensity is also evident in the Fourier-transformed to k-space EXAFS spectra after the third-shell filtering (compare Figs. 3c and 4c). However, there is practically no change in the third-shell intensity in CH₂Br₂. In CBr₄ the third-shell peak maximum shifts to longer distances (see Figs. 2, 3c, and 4c), whereas in CHBr₃ and CH₂Br₂ the position of the peak remains the same. The third-shell single scattering path fitting was successful only in CBr₄, where the average intramolecular Br-Br distance increases by 0.06 Å, and the coordination number decreases by 1.6 in comparison with the ones found for the nonirradiated sample. In irradiated CHBr₃, CH₂Br₂, and nonirradiated CH₂Br₂ (see above), the third-shell signal contains contribution from some additional single or multiple scattering paths. However, the signal-to-noise ratio of the experimental data does not allow us to determine reliably the structure of these paths. We believe that bromine photoreduction process is still present in CH₂Br₂. The third shell, however, becomes insensitive to decreasing Br-Br intramolecular coordination number, because a major part of the signal in the shell is strongly influenced by other scattering paths.

All irradiated samples also show change in the second shell, which is attributed to intermolecular Br-Br single scattering. As discussed above, in the nonirradiated molecules we observed increase in the second-shell coordination number with decrease in the number of bromines per molecule. Since UV irradiation causes bromine atoms to dissociate from the molecules, we expect to find more bromines present in the second shell. In fact this is the case in CBr₄ and CHBr₃ molecules: the second shell coordination number increased by 2.0 and 0.7, respectively. However, there is a strong decrease in the coordination number of the second shell in CH₂Br₂. Fig. 2 shows that the second-shell maximum shifts to shorter distances in all three samples. In CBr₄ and CHBr₃ molecules this shift is observed within our experimental error, whereas in CH₂Br₂ it is much larger, 0.11 Å.

There is practically no change in the first shell in all of the molecules, except a small shift toward shorter distances in CBr₄ and CHBr₃ and toward longer distances in CH₂Br₂. However, these shifts are not severe and also fall within the experimental error.

The above data show that there is a significant change in EXAFS spectra before and after irradiation. We have explained the main features of these changes; however, we did not determine the structure of all photodisociation products. The reason for this is that continuous irradiation of the same sample causes the dissociation of the product with subsequent association of the radical and other intermediates formed during photolysis. Long-term photolysis generates a number of products with diverse structures, which complicates the structure determination based on EXAFS data. Additional experiments in flowing samples, where only the primary dissociation and recombination take place, are needed. The experimental data and processing will be suitable for an accurate structure determination of the primary intermediates and products in solution with picosecond time resolution and 0.02-Å bond length accuracy.

The ultrafast EXAFS experimental system also can be used to investigate the evolution of short-lived structures of intermediates that evolve during the course of biological reactions. The structures of the photochemical intermediates and products created during the course of the binding reaction of copper aminothiocarbonates to melanine (44) can be studied by means of both ultrafast laser spectroscopy and ultrafast EXAFS spectroscopy. Such experiments can provide the data for the elucidation of the binding mechanism and, in addition, knowledge of the changes in the coordination number of the metal of the short-lived intermediates and products. Such studies would help to guide the design of new materials that will be more effective in binding and reacting with specific biological agents, such as melanine and other viruses. The fact that these studies are performed in solution allows the determination of the structures emerging during the course of real-life biological reactions.

Abbreviations

EXAFS

extended x-ray absorption fine structure

CBr₄

carbon tetrabromide

CHBr₃

bromoform

CH₂Br₂

dibromomethane