Reactions of Atomic Hydrogen with Formic Acid and Carbon Monoxide

Aug 12, 2014 - Department of Chemistry, University of Wyoming, Laramie, ... Leif O. Paulson , Fredrick M. Mutunga , Shelby E. Follett , and David T. A...
1 downloads 0 Views 2MB Size
Download PDF
Article pubs.acs.org/JPCA

Reactions of Atomic Hydrogen with Formic Acid and Carbon Monoxide in Solid Parahydrogen II: Deuterated Reaction Studies William R. Wonderly and David T. Anderson* Department of Chemistry, University of Wyoming, Laramie, Wyoming 82071, United States S Supporting Information *

ABSTRACT: It is difficult to determine whether the measured rate constant for reaction of atomic hydrogen with formic acid reported in Part 1 reflects the H atom quantum diffusion rate or the rate constant for the tunneling reaction step. In Part 2 of this series, we present kinetic studies of the postphotolysis H atom reactions with deuterated formic acid (DCOOD) to address this ambiguity. Short duration 193 nm in situ photolysis of DCOOD trapped in solid parahydrogen results in partial depletion of the DCOOD precursor and photoproduction of primarily CO, CO2, DOCO, HCO and mobile H atoms. At 1.9 K we observe post-irradiation growth in the concentrations of DOCO and HCO that can be explained by H atom tunneling reactions with DCOOD and CO, respectively. Conducting experiments with different deuterium isotopomers of formic acid (DCOOD, DCOOH, HCOOD and HCOOH) provides strong circumstantial evidence the reaction involves H atom abstraction from the alkyl group of formic acid. Further, the anomalous temperature dependence measured for the H + HCOOH reaction in Part 1 is also observed for the analogous reactions with deuterated formic acid. The rate constants extracted for H atom reactions with DCOOD and HCOOH are equivalent to within experimental uncertainty. This lack of a kinetic isotope effect in the measured rate constant is interpreted as evidence the reactions are diffusion limited; the measured rate constant reflects the H atom diffusion rate and not the tunneling reaction rate. Whether or not H atom reactions with chemical species in solid parahydrogen are diffusion limited is one of the outstanding questions in this field, and this work makes significant strides toward showing the reaction kinetics with formic acid are diffusion limited.

1. INTRODUCTION One of the first research groups to study pure tunneling reactions focused on creating H atoms in solid molecular hydrogen (SMH).1−22 These are fascinating chemical systems due to the importance of nuclear quantum effects; the light mass of the H atom results in a large de Broglie wavelength, and the H2 molecule undergoes both large amplitude translational zero-point motions and free rotation within SMH.23 Through a series of radiolysis and photolysis studies of solid H2−D2 mixtures and solid HD, Miyazaki and his group (followed by Kumada) used electron spin resonance (ESR) spectroscopy to demonstrate a number of nonstandard chemical phenomena: (1) H atoms readily diffuse through SMH even at temperatures below 5 K by repeated H + H2 → H2 + H tunneling exchange reactions, (2) D atoms produced in solid H2 rapidly react with the host thereby trapping the D atom in HD and creating mobile H atoms, and (3) H atoms in solid H2 are remarkably stable on the time scale of hours due presumably to inefficient H + H → H2 recombination in enriched parahydrogen crystals.1−22 Proof of the importance of tunneling in the measured reaction dynamics10 is provided by comparison of the rate constant for the D + HD → H + D2 reaction estimated experimentally to be 3.5 × 10−27 cm3 molecule−1 s−1 at 4.2 K with the thermally activated rate constant calculated to be 10−400 cm3 molecule−1 s−1. Further © 2014 American Chemical Society

proof the reaction occurs exclusively through tunneling is provided by measuring the rate constants for different isotopic variants of this reaction. The 3 × 104 enhancement7 in the rate constant for D + H2 → HD + H compared to H + D2 → HD + D is an example of a large kinetic isotope effect (KIE) that can occur in tunneling dominated reactions when the effective mass for the reaction coordinate is changed. Thus, the H + D2 reaction has a much smaller rate constant because a heavier D atom is abstracted, while in the other reaction an H atom is abstracted. In addition, the Born−Oppenheimer reactive potential energy surface for both reactions are rigorously thermoneutral; however, differences in the asymptotic vibrational zero-point energies of both reactants and products are the reason such large KIEs are expected. In the present example, the vibrationally adiabatic ground-state (VAG) potential reaction D + H2 → HD + H is exothermic (−260 cm−1) while the VAG for H + D2 → HD + D is endothermic (+320 cm−1). At low enough temperatures, therefore, the H + D2 reaction cannot occur even by tunneling because a chemical system cannot tunnel uphill. This highlights one of the reasons Received: March 11, 2014 Revised: August 6, 2014 Published: August 12, 2014 7653

dx.doi.org/10.1021/jp502469p | J. Phys. Chem. A 2014, 118, 7653−7662

The Journal of Physical Chemistry A

Article

Table 1. Peak Positions (FWHM) of trans-DCOOD, trans-DCOOH, and trans-HCOOD Isolated in Solid pH2a sym A′ A′ A′ A′ A′ A′ A′ A′ A″ A′ A′ A′ A′ A″ A′ A′ A′ A′ A′ A′ A′ A″

mode trans-DCOOD ν1 (O−D str) ν2 (C-D str) ν4 + ν5 2ν8 ν3 (CO str) ν4 (C−O str) ν5 (DCO−COD) ν6 (COD-DCO) ν8 (CD wag) trans-DCOOH ν2 (C-D stretch) ν3 (CO stretch) ν5 (C−O stretch) ν6 (CD rock) ν8 δ(CD wag) trans-HCOOD ν1 (C−H stretch) ν3+ν5 ν2 ν(O−D) ν3 ν(CO) ν4 (CH rock) ν5 ν(C−O) ν6 (CD rock) ν8 (CH wag)

gas

Nel

pH2

m

2631.87b 2232c

2632.1 2192.5 1761.2 1725.5 1170.5 1031.9 944.9 874.8

2621.71(mult) 2231.8(1.07) 2196.1(2.02) 1758.86(0.21) 1720.87(0.44) 1172.45(mult) 1040.3(1.41) 947.22(1.33) 874.5(0.36)

0.9020(91) 0.5054(22) 0.2501(12)

2219.69f 1725.87g 1142.31h 970c 870c

2215.5 1725.47 1140.8

2219.8(1.7) 1723.09(0.11) 1143.1(mult)

874.8

859.4

3.385(11) 1.000 ----

2944c

2939.4

2633.5c 1772.12i

2631.6 1770.9 1368.1 1177.0 971.1 1037.4

2952.41(mult) 2935.55(1.3) 2621.55(mult) 1765.85(mult) 1368.51(mult) 1178.79(mult) 975.6(4.6) 1037.94(0.10)

0.053(10) 0.03532(38) 0.2004(35) 1.0000 0.01002(14) 0.6934(49) ---0.02121(9)

1742c 1725.12d 1170.80e 1040c 945c 873c

1177.09j 972.85k ----

0.1232(4) 1.0000 0.0300(3) 0.6828(44)

a f

The m values are the slopes determined in intensity−intensity correlation plots. bReference 47. cReference 32. dReference 33. eReference 48. Reference 49. gReference 50. hReference 51. iReference 35. jReference 52. kReference 53. lReference 32.

purely experimental evidence that the reaction is occurring via tunneling. In the companion paper30 (hereafter referred to as Part 1), we describe our experimental studies of the H + HCOOH → H2 + HOCO reaction in solid parahydrogen (pH2) over the temperature range from 1.7 to 4.3 K. To study this lowtemperature reaction, we deposit HCOOH-doped pH2 solids and utilize 193 nm in situ photolysis of HCOOH to generate H atoms for the subsequent chemical reactions. We set up the photolysis conditions such that only a portion of the HCOOH precursor is photolyzed so that after photolysis is stopped, there is a significant amount of HCOOH available for reaction. This differs from the earlier tunneling reaction studies of Momose in that the H atoms generated during photolysis must diffuse to find potential HCOOH reactants. In the CD3 + H2 tunneling reaction studies of Momose and co-workers,25−27 the generated CD3 does not have to diffuse to the other reactant because it reacts directly with the pH2 host. In this way our experiments extend the first studies of H atom reactions to study a number of H atom tunneling reactions with species other than molecular hydrogen. In Part 2 of this series, we explore the analogous H atom reactions starting with a DCOOD precursor to test the argument put forth in Part 1 that the low temperature H + HCOOH → H2 + HOCO tunneling reaction is responsible for the growth in HOCO and to measure the rate constant for the analogous H + DCOOD → HD + DOCO reaction. The first point is important because for spectroscopic reasons we cannot measure the decay in the HCOOH concentration due to reaction to help support the claim we are studying the reaction of H atoms with HCOOH. Therefore, if we detect a strong dependence of the branching between DOCO and HOCO on

why chemistry governed by quantum control can be very different from the chemistry governed by kinetic control.24 The first group to extend these types of tunneling reaction studies in SMH to chemical reagents other than hydrogen was the Momose group.25−27 For example, in a series of experiments they studied the CH3 + H2 → CH4 + H reaction, for all possible isotopomers of CH3, using Fourier transform infrared (FTIR) spectroscopy by generating CH3I-doped parahydrogen (pH2) solids and using in situ UV photolysis to generate the CH3 reagent. These were important direct spectral measurements on reactions between neutral molecules that occur exclusively by tunneling.25−27 They showed that CD3 decayed in solid pH2 at 5 K with a first-order rate constant of k = 3.3 × 10−6 s−1 (τ = 1/k = 3.5 days) while CH3 did not noticeably decay over a week’s time.25−27 The authors attributed this apparently large KIE to zero-point vibrational energy differences, which make the VAG involving CH3 slightly endothermic and thus not possible via quantum mechanical tunneling at 5 K. However, subsequent theoretical studies showed the VAG for both reactions are exothermic, but the barrier height and width of the VAG are smaller for CD3 + H2 compared with CH3 + H2, thereby qualitatively explaining the KIE ≥ 41 experimental results.28 Given the density of solid pH2,23 this low temperature rate constant for CD3 + H2 corresponds to a bimolecular rate constant of k = 1.4 x10−28 cm3 molecule−1 s−1, which is many orders of magnitude larger than the k = 1.8 × 10−431 cm3 molecule−1 s−1 rate constant predicted from the most recent recommended Arrhenius expression for the 200−2000 K temperature range.29 This clearly demonstrates that at this low temperature the reaction occurs exclusively by tunneling. In addition, the qualitative differences in the reactivity of CH3 and CD3 provide further 7654

dx.doi.org/10.1021/jp502469p | J. Phys. Chem. A 2014, 118, 7653−7662

The Journal of Physical Chemistry A

Article

shown in Figure 1. The three peaks shown in Figure1b are all relatively strong absorptions, and the ν3 carbonyl stretching

the H/D composition of the formic acid isotopomer, we can provide support that formic acid is one of the reactants. We also expected to measure a large KIE for the H + DCOOD → HD + DOCO reaction because it involves abstraction of a D atom instead of an H atom, but as we will show, we measure nearly equivalent rate constants. This strongly suggests the reaction kinetics is diffusion limited: the extracted rate constant reflects the H atom diffusion coefficient in solid pH2 and not the rate of the tunneling reaction step. Consistent with the findings presented in Part 1, the reaction kinetics at both 1.9 and 4.3 K are qualitatively different, and we only observe DOCO growth at low temperature.30 Characterization of this inverse temperature effect on the measured reaction kinetics is an important step forward in our understanding of H atom tunneling reactions in solid pH2 and proves the statement in the earlier 2004 paper of Momose and co-workers25 prophetic: “On the other hand, temperature dependence on the tunneling rate has not yet been studied, but it should be. This dependence will give us important information on the effect of fluctuations of the environment on the tunneling process. Solid parahydrogen will provide a useful matrix for such studies.”

2. EXPERIMENTAL AND RESULTS Our experimental apparatus and basic procedures are described in Part 1.30 In the present studies we use fully deuterated formic acid (DCOOD) as a dopant. The DCOOD sample was obtained from Sigma-Aldrich and used as delivered after several freeze−pump−thaw cycles. The concentrations of various chemical species studied in this work are determined using the Beer’s Law expression in Part 1 and the integrated absorption coefficients listed in Table S1 provided in the Supporting Information. The accuracy of the reported absolute concentrations likely varies for each molecule studied, but we estimate the error in the concentrations of DCOOD to be on the order of ±50% based on the concentrations determined using the ν1 and ν2 absorptions. 2.1. IR Spectroscopy of Deuterated Formic Acid. We first characterize and assign the IR spectrum of DCOOD isolated in solid pH2. In previous studies of HCOOH/pH2, we made the assignments based on comparison of the measured peak frequencies to literature values.31 In the present studies of DCOOD, isotopic scrambling can occur prior to photolysis leading to detectable concentrations of DCOOH, HCOOD, and HCOOH that can complicate assignments. Therefore, a series of spectra with varying peak intensities recorded during the deposition of 4 separate DCOOD/pH2 experiments are used to construct intensity versus intensity correlation plots. All the peaks assigned to a given isotopomer necessarily must show strong linear intensity correlations (I = mIref). Variations in the isotopomer distribution for the DCOOD sample held in the dopant manifold for over a week allows correlation plots to distinguish peaks for different isotopomers. Further, a fifth experiment was conducted using a dopant made by adding an aliquot of HCOOH to the stainless steel sample holder that previously contained DCOOD, creating a dopant sample that contains HCOOH, but also significant amounts of HCOOD. The binary intensity correlation plots for DCOOD, DCOOH, and HCOOD used to confirm the vibrational assignments are presented Figures S1−S3 in the Supporting Information. The resulting DCOOD, DCOOH, and HCOOD peak assignments are presented in Table 1 along with literature values. Representative spectra for all the peaks assigned to transDCOOD (hereafter simply DCOOD) isolated in solid pH2 are

Figure 1. Survey spectrum showing all the peaks assigned to transDCOOD isolated in solid pH2. The spectra in parts a and c are recorded at 1.92 K for a 0.25(1) cm thick as-deposited sample. The spectrum in part b is recorded during deposition at 2.2 K for the same sample when it was only 0.04(2) cm thick to keep these stronger peaks on-scale. The ν3 peaks of HCOOD and HCOOH are labeled accordingly.

mode of DCOOD is strongly perturbed by a Fermi resonance with the 2ν8 state of the out-of-plane CD wag.32,33 This Fermi interaction results in the 2ν8 transition being pushed up in energy while ν3 is pushed down. Note in Figure 1b the ν3 peaks of HCOOH and HCOOD are labeled, and missing from this progression are the ν3 peaks of DCOOH and DCOOD, which are pushed down in energy by the Fermi resonance with 2ν8. Furthermore, the 2ν8 overtone steals intensity from the ν3 fundamental such that under these conditions (initial DCOOD concentration of 57 ppm and d = 0.25(1) cm), both the ν3 and 2ν8 peaks are saturated and cannot be used to monitor the DCOOD concentration. Indeed, the spectra shown in Figure 1b were recorded during deposition while the sample was still quite thin (e.g., d = 0.04 cm). For most of the kinetic measurements, we use the InSb detector, which has a cutoff below 1800 cm−1; therefore we will use the ν2 (2231.8 cm−1) and ν1 (2621.71 cm−1) peaks to monitor the DCOOD concentration in the sample. We also must characterize the amounts of DCOOH, HCOOD, and HCOOH present in the sample before and after photolysis. This is easy for HCOOH because we previously assigned this spectrum and use the ν2 peak at 2942.07 cm−1 to monitor the HCOOH concentration.31 However, we must assign the DCOOH and HCOOD features because the spectra of these two isotopomers have not been reported in solid pH2. Fortunately, the peak positions for 7655

dx.doi.org/10.1021/jp502469p | J. Phys. Chem. A 2014, 118, 7653−7662

The Journal of Physical Chemistry A

Article

assigned to ν3 (CO stretch), a multiplet feature at 1160.78 cm−1 assigned to ν5 (C−O stretch), and a peak at 2672.5 cm−1 assigned to ν2 (O−D stretch). Representative spectra of the peaks assigned to cis-HCOOD are presented in Figure S4 (Supporting Information). These assignments are based on the proximity of these features to the peak positions reported for nH2 matrices34 and the predicted 348.2, 434.2, and 128.6 km mol−1 intensities of the ν5, ν3, and ν2 vibrational modes, respectively.36 2.2. 193 nm Photoinduced DCOOD Chemistry. From now on when we refer to the different formic acid isotopomers, we mean the lower energy trans conformation unless explicitly stated otherwise. Shown in Figure 3 are three spectra in the

HCOOD have been reported for a normal-hydrogen (nH2, 75% oH2, and 25% pH2) matrix at 4.2 K, and we can use these assignments as a guide.34 Using the same approach as with DCOOD, we can definitely assign seven peaks to transHCOOD and three to trans-DCOOH (hereafter simply HCOOD and DCOOH). The peak assignments and frequencies are presented in Table 1. We show representative absorptions for HCOOD and DCOOH in Figure 2. Using

Figure 2. Representative spectra of peaks assigned to trans-DCOOH and trans-HCOOD isolated in as-deposited solid pH2 at 1.85 K. The peak marked with an asterisk is the ν3 peak of trans-HCOOH.

experiments with different relative isotopomer concentrations (DCOOD versus HCOOD), we find that the ν2 peak (O−D stretch) of HCOOD overlaps with the ν1 peak (O−D stretch) of DCOOD, and this is consistent with the greater scatter in the DCOOD ν1 intensity correlation data for this peak (see Figure S1). The ν3 absorption feature of HCOOD is not perturbed by a strong Fermi resonance with the 2ν8 mode and shows multiple peaks similar to the analogous ν3 absorption in HCOOH that was assigned to orthohydrogen (oH2) clustering to the dopant.31 The ν3 feature for HCOOD is shown in Figure 2 along with the HCOOH ν3 feature at slightly higher wavenumbers (marked with an asterisk). Upon closer inspection of this HCOOD ν3 feature, however, the pattern of the multiplets is more complicated: there are two overlapping progressions of peaks to lower energy. This more complicated multiplet pattern is likely caused by a Fermi interaction of the ν3 mode with the ν5 + ν7 combination mode as identified in high-resolution gas phase studies of the ν3 fundamental.35 As discussed above for DCOOD, the DCOOH isotopomer also shows a strong Fermi interaction between the ν3 and 2ν8 vibrational modes. In this work we use the ν1 absorption for HCOOD and the ν2 absorption for DCOOH to monitor the concentrations of these two species. Finally, we tentatively assigned peaks to the higher energy cis conformer of HCOOD using the peak frequencies reported for nH2 matrices.34 Deuteration of the OH group increases the lifetime of cis-HCOOD by several orders of magnitude such that if cis-HCOOD is produced during UV photolysis it should be metastable. We observe a number of peaks that are weak in the spectra of the as-deposited sample but significantly increase in intensity after photolysis that we tentatively assign to cisHCOOD; we observe peaks at 1797.25 and 1817.26 cm−1

Figure 3. IR absorption spectra in the DCOOD ν2 and ν1 regions for a DCOOD-doped pH2 sample showing the effects of 193 nm photolysis and time. Trace (a) is recorded at 1.92 K before photolysis, trace (b) is recorded with a 4.8 min acquisition time (16 scans at 0.03 cm−1) immediately after 7.2 min of photolysis (85 μJ cm−2 pulse−1, 108,000 pulses, 250 Hz), and trace (c) is recorded 568 min after trace (b) while maintaining the sample at 1.91(2) K. The sample is 0.25(1) cm thick with [DCOOD] = 57 ppm, [DCOOH] = 8 ppm, [HCOOD] = 2 ppm, and [HCOOH] = 0.5 ppm prior to photolysis. Spectra are offset vertically for ease of comparison.

regions of the ν2 and ν1 DCOOD peaks recorded during the course of a 193 nm photochemical experiment on a DCOOD/ pH2 sample. Trace (a) in Figure 3 is recorded at 1.92 K approximately 39 min after deposition and before photolysis. We estimate the sample contains 57 ppm of DCOOD, 7.6 ppm of DCOOH, 1.7 ppm of HCOOD, and 0.5 ppm of HCOOH prior to photolysis. Shown in traces (b) and (c) in Figure 3 are rapid scan FTIR spectra (289 s acquisition times, 16 scans, 0.03 cm−1 resolution) recorded right after photolysis (85 μJ cm−2 pulse−1, 108,000 pulses, 250 Hz) and 568 min after photolysis, respectively, while holding the sample at 1.91(2) K. As can be seen in Figure 3b, the DCOOD peak intensity is reduced, but the sample still contains approximately 68% of the initial DCOOD concentration after photolysis. As in the photochemical experiments on HCOOH/pH2 reported in Part 1, there are significant changes in the lineshapes of the two 7656

dx.doi.org/10.1021/jp502469p | J. Phys. Chem. A 2014, 118, 7653−7662

The Journal of Physical Chemistry A

Article

through the solid. The D atoms cannot move via exchange tunneling because the D atom gets trapped21 in an HD molecule after the first D + H2 → HD + H exchange reaction. This isotopic variant of the H atom exchange reaction is exothermic (−260 cm−1) due to the lower zero-point vibrational energy of HD compared to H2 and readily occurs with a time constant of 77 s at 4.2 K in solid D2-H2 mixtures.9 Accordingly, the growth in the HCO peak after photolysis can be explained by the H + CO → HCO reaction. The growth in the DOCO concentration after photolysis is therefore taken as strong indirect evidence for the reaction H + DCOOD → HD + DOCO. Unfortunately, single or double IR transitions of HD in solid pH2 are too weak to be used to detect these species under these conditions.38 Presumably, the reaction of H atoms with DCOOH leads to the slight production of HOCO shown in Figure 4. As we will show, the production of DOCO versus HOCO is strongly correlated to the isotopic makeup of the formic acid precursor. The carrier of the weak (but reproducible) unlabeled features in Figure 4 is thought to be HCO clustered with other species. While we did not attempt to definitively assign these tentative HCO cluster features (HCO− D2O, HCO−HDO, HCO−DCOOD), they are not due37 to DCO or the higher energy cis conformers of DOCO, HOCO, or DCOOD. 2.3. H + DCOOD Reaction Kinetics at 1.9 K. Shown in Figure 5 is the complete kinetic data set for this DCOOD/pH2

DCOOD features shown in Figure 3 between traces (b) and (c). Rigorously, the absorption features shown in Figure 3 are due to DCOOD(oH2)n clusters that form during deposition, and we speculate that the changes in the ν2 and ν1 lineshapes are due to intracluster oH2 to pH2 nuclear spin conversion similar to the results presented in Part 1.30 However, unlike HCOOH(oH2)n we cannot characterize the intracluster nuclear spin conversion process using the ν3 or 2ν3 absorptions to test this hypothesis because the ν3 vibrational mode of DCOOD is severely perturbed by Fermi interactions with 2ν8. Nonetheless, these small changes in the lineshapes of the ν2 and ν1 peaks shown in Figure 3 make quantifying small DCOOD concentration differences problematic, and we will discuss this in more detail shortly. The same three spectra shown in Figure 3 for a photolysis experiment on a DCOOD/pH2 sample are shown in the 1840 to 1870 cm−1 region in Figure 4. We observe small peaks that

Figure 4. FTIR absorption spectra from 1840 to 1870 cm−1 for the same spectra described in Figure 3 displaying continued growth of DOCO, HOCO, and HCO after photolysis. The additional unlabeled peaks in this region are likely due to HCO clustered with other species but were not assigned. Spectra are offset vertically for ease of comparison.

can be assigned to trans-DOCO, trans-HOCO, and HCO right after photolysis (hereafter referred to as DOCO and HOCO). The assigned HOCO and HCO peaks are reported in Part 1, and the DOCO peak assignments are reported in Table S2 in the Supporting Information. Similar to the photochemical studies on HCOOH/pH2 reported in Part 1, these three absorption features steadily increase in intensity while the sample is held at 1.91(2) K after photolysis. No DCO peaks37 were detected in these experiments either directly after photolysis or in the repeated FTIR scans after photolysis. We speculate that DCO produced during photolysis of DCOOD undergoes efficient secondary photolysis such that under these conditions no peaks due DCO are observed. We only observe growth of species after photolysis that can be attributed to H atom reactions with DCOOD, DCOOH, and CO. Our results are therefore consistent with the earlier H atom reaction studies21 that showed that D atoms are not mobile in solid H2 due to the unique manner in which these species diffuse

Figure 5. Kinetic plots of concentration versus time recorded before and after photolysis for the DCOOD/pH2 sample depicted in Figures 3 and 4. The photolysis interval is indicated by a gray vertical bar. The data are represented by dotted circles and the lines are the result of least-squares fits to the respective data (see text for details). The arrows indicate the three spectra shown in Figures 3 and 4 during the course of the full kinetic experiment.

photolysis experiment; the three spectra shown in Figures 3 and 4 from this experiment are indicated by arrows in Figure 5. The gray vertical bar represents the timing and duration of the 193 nm photolysis exposure. In order to show the relative concentrations of DCOOD, DCOOH, HCOOD, CO, and CO2 before and after photolysis, we convert the measured 7657

dx.doi.org/10.1021/jp502469p | J. Phys. Chem. A 2014, 118, 7653−7662

The Journal of Physical Chemistry A

Article

Previous studies31,39 have documented that when HCOOH is deposited in highly enriched pH2 solids, it forms clusters with residual oH2 molecules present in the solid during deposition or upon annealing. Therefore, we speculate the spectra shown here in Figures 1 and 3 are due to DCOOD(oH2)n clusters in solid pH2, where n represents of the number of oH2 molecules in the cluster. Previously,31 we could assign the specific n values for HCOOH(oH2)n clusters using the ν3 fundamental, which shows resolved peaks with nearly a constant 0.38(6) cm−1 shift in the peak position of each cluster to lower wavenumbers with increasing n. However, for DCOOD this is not possible because the ν3 vibration displays a strong Fermi resonance with 2ν8, thereby changing the character of this vibrational mode, and no resolved cluster peaks are observed. The cluster distribution cannot be measured using other fundamental vibrations because the other modes do not lead to spectrally resolved ncluster peaks. This means that while we suspect the apparent linear decrease in the DCCOD concentration shown in Figure 5 is caused by slow intracluster oH2 to pH2 nuclear spin conversion (NSC), we cannot prove this by measuring how the DCOOD(oH2)n cluster distribution changes with time like we could for HCOOH(oH2)n. Thus, we are left to speculate that during the course of the 10 h reaction study shown in Figure 5 there is significant intracluster NSC to change the DCOOD(oH2)n cluster distribution, and this subtly alters both the ν1 and ν2 lineshapes. These changes in the ν1 and ν2 lineshapes are observed (see Figure 3), we just do not have an independent measurement of the cluster distribution. We further speculate these changes in the DCOOD lineshapes effectively mask the small concentration changes (∼0.3 ppm) that occur due to reaction with H atoms. Further support for this interpretation of the data will be shown in the following section when we study the products of reaction for a sample containing primarily HCOOH and HCOOD. As can be seen in Figure 5, the growth in HCO is markedly biexponential with both slow and fast components. This is consistent with the previous photolysis experiments using HCOOH under conditions where [CO] ≈ [HCOOH]. Therefore, in this case, the HCO data are fit to a biexponential expression,

concentrations (in ppm) to mole fractions by dividing the concentration of each species by the sum of all five species. Under these photolysis conditions, the DCOOD mole fraction is reduced from 0.86 before photolysis to 0.60 after; however, DCOOD still has the largest mole fraction of any of the five species. Consistent with the kinetic measurements on HCOOH doped pH2 samples, the DCOOD, CO and CO2 concentrations remain constant after photolysis with significantly more CO (0.28) produced than CO2 (0.02). We can fit the kinetic data for DOCO and HOCO to a single exponential expression, ΔC(t ) = C0 + C1(1 − exp(−k1t ))

(2)

where t is the time after photolysis, C0 and C0 + C1 are the concentrations at time zero and at infinite reaction time, and k is the first-order rate constant. In this analysis, we define time zero as when the photolysis is stopped. We fit the data shown in Figure 5 and extract rate constants for the growth of DOCO and HOCO of k = 3.44(6) × 10−3 min−1 and k = 2.9(4)x10−3 min−1, respectively. The resulting fits are shown as solid lines in Figure 5, and the parameters extracted from the fit are presented in Table 2. Consistent with the assumed reaction Table 2. Kinetic Parameters Determined from Least Squares Fits to Eqs 2 and 3a [HCO]0 [HCO]1 k1/10−3 min−1 [HCO]2 k2/10−3 min−1 [DOCO]0 [DOCO]1 k1/10−3 min−1 [HOCO]0 [HOCO]1 k1/10−3 min−1

expt. 1

expt. 2

expt. 3

0.0165(31) 0.154(15) 28.0(29) 0.197(14) 6.36(49) 0.023(12) 0.3299(21) 3.436(61) 0.0079(84) 0.038(20) 2.94(39)

0.0167(27) 0.166(42) 28.7(55) 0.399(93) 3.5(22) −0.0034(18) 0.3088(62) 6.61(29) 0.0053(11) 0.047(59) 1.8(28)

0.1127(42) 0.0365(54) 82(25) 0.1242(31) 6.98(93) 0.0111(26) 0.556(67) 2.17(35) 0.0537(8) 0.5675(53) 4.036(67)

a

All concentrations in ppm. The uncertainties in parentheses represent 1σ from the least-squares fit.

scheme, the two reactions that lead to DOCO and HOCO formation involve D atom abstraction from DCOOD and DCOOH, respectively, and thus both should have comparable rate constants as observed experimentally. Therefore, we also expect the branching between DOCO and HOCO should scale roughly with the concentrations of DCOOD and DCOOH after photolysis. Consistent with this expectation, in this experiment the mole fraction of DCOOD is 7.1 times larger than DCOOH, and the branching between DOCO and HOCO is 8.6. As discussed in section 2.2, the fact that we observe growth in the DOCO signal after photolysis of a DCOOD/pH2 sample is strong evidence for the H + DCOOD → HD + DOCO reaction. We therefore should observe an analogous exponential decrease in the DCOOD concentration after photolysis if we are observing a single elementary step. However, as shown in Figure 5, we do not observe this exponential decay in the DCOOD mole fraction but rather a small linear decrease over the entire 568 min observation time after photolysis. We observed a similar nearly linear decrease in the HCOOH mole fraction after photolysis in our analogous studies of H + HCOOH reported in Part 1 of this series.30 We believe the reason for this apparent discrepancy is the same in both cases.

ΔC(t ) = C0 + C1(1 − exp(−k1t )) + C2(1 − exp(−k 2t )) (3)

with both slow and fast rate constants. The results of the fit are shown as a solid red line in Figure 5 and the fitted parameters are presented in Table 2. For the data shown in Figure 5, fits to the HCO data result in k1 = 6.4(5) × 10−3 min−1 and k2 = 28(3) × 10−3 min−1. This is expected because even in the photolysis of DCOOD, it is H atoms generated as byproducts of reactions of photoproducts (D + H2 → HD + H and OD + H2 → HDO + H) with the pH2 host that are the mobile reactive species that mediate the chemistry. The H atom reactions that lead to HCO growth therefore compete with the H + DCOOD and H + DCOOH reactions in the decay of the H atom species. For the experiment depicted in Figure 5, we estimate the ratio of the sum of DCOOD and DCOOH mole fractions to CO is approximately 2.4, whereas the ratio of the sum of DOCO and HOCO concentrations to HCO is only 1.1. This indicates there is a slight preference for H atom reactions with CO compared to either formic acid isotopomer. Further, comparison of the kinetic data in Figure 5 with similar plots for H atom reactions with HCOOH (for example, see Figure 4 in Part 1) shows a greater HCO branching ratio for reactions 7658

dx.doi.org/10.1021/jp502469p | J. Phys. Chem. A 2014, 118, 7653−7662

The Journal of Physical Chemistry A

Article

to 1.91(1) K, and three successive FTIR spectra are recorded. No noticeable change in the concentrations of HCO, DCOOD, or DCOOH is observed during this cool down period, suggesting that the concentration of H atoms at this point of the experiment is negligible. Then the same sample is photolyzed for a second time (115 μJ cm−2 pulse−1, 108,000 pulses, 250 Hz) with two FTIR scans recorded during photolysis, and after photolysis FTIR scans are recorded every 12 min until the cryostat ran out of liquid helium. This second photolysis depletes a comparable amount of DCOOD and also the ∼0.1 ppm of HCO produced in the first photolysis. Once again, under these low temperature conditions we observe growth in both the DOCO and HCO concentrations similar to the other low temperature experiment. The rate constant extracted from the DOCO growth (k = 6.6(3) × 10−3 min−1) is slightly greater but comparable to the rate constant extracted from the data in Figure 5. Once again, HCO shows biexponential growth with rate constants (k1 = 3.5(22) × 10−3 min−1 and k2 = 29(6) × 10−3 min−1) comparable to the earlier low temperature experiment. Comparing the two photoinduced kinetic experiments shown in Figure 6 performed on the same sample, the first experiment did not produce significant amounts of DOCO or HCO while the second low temperature photolysis did initiate postphotolysis reactions that lead to the formation of HCO and DOCO (HOCO). This clear inverse temperature behavior is consistent with the measurements on the H + HCOOH reaction presented in Part 1 and cannot be explained by standard low temperature Wigner threshold laws for a single elementary reaction step.40,41 We interpret this as evidence the reaction kinetics involve more than one step, and this will be explored further in the Discussion section. To investigate these changes with temperature further we performed another experiment using a sample that contained mostly HCOOH (0.70), but also HCOOD (0.23) and practically no DCOOH (0.03) or DCOOD (