Letter pubs.acs.org/JPCL
Automated Exploration of Photolytic Channels of HCOOH: Conformational Memory via Excited-State Roaming Satoshi Maeda,*,† Tetsuya Taketsugu,† and Keiji Morokuma*,‡,§ †
Department of Chemistry, Faculty of Science, Hokkaido University, Sapporo 060-0810, Japan Fukui Institute for Fundamental Chemistry, Kyoto University, Kyoto 606-8103, Japan § Department of Chemistry and Cherry L. Emerson Center for Scientific Computation, Emory University, Atlanta, Georgia 30322, United States ‡
S Supporting Information *
ABSTRACT: To elucidate the photodissociation mechanism of HCOOH, we systematically explored reaction pathways starting from the first excited singlet state (S1) by using automated reaction path search methods. All critical points, that is, minima, transition states, minimum energy conical intersections, and minima on seam of crossing, for the S0, T1, and S1 potential energy surfaces (PESs) obtained in the present search were optimized at the CASPT2 level. The structure list obtained by the search explained all experimentally reported photolytic channels. A new mechanism for the previously suggested but unexplained conformational memory in the 193 nm photolysis is proposed, which involves two steps: partial dissociation and succeeding roaming of one of H atoms on the S1 PES, followed by intramolecular recombination on the S0 PES after radiationless transition through a conical intersection at a partially dissociated geometry. This is partially similar to the excited-state roaming recently discovered for the NO3 radical. SECTION: Spectroscopy, Photochemistry, and Excited States
M
depending on the initial conformation was seen in a trajectory surface hopping (TSH) simulation on PESs of reparameterized semiempirical AM1 theory.8 Although the direction of this change is consistent with the experiment, the magnitude was too small to be compared with the experiment. The authors suggested two reasons regarding the discrepancy: a strong cage effect of the Ar matrix and a lack of important mechanism(s) in the reparameterization procedure of the AM1 theory. Nevertheless, the cage effect on the cis−trans isomerization dynamics cannot be considered to be critical because the cis−trans isomerization barrier in the Ar matrix is comparable to that in the gas phase.9 Very small cage effects on the cis−trans isomerization were also shown by theoretical calculations.7,9 Therefore, systematic exploration of unknown mechanisms is required. Very recently, excited-state roaming was predicted in the photolysis of the nitrate radical (NO3).10,11 Although the roaming mechanism has been discovered in photolysis and pyrolysis of a number of small molecules,12−17 all of them have been found to occur on the ground-state PESs, even in photolysis. The excited-state roaming mechanism of NO3 photolysis was able to explain previous experimental observations18,19 and was very recently confirmed by ion-imaging experiments.20,21 However, a question remains: is such an
olecular conformation is one of the most significant research subjects, including protein folding, molecular motor, photoinduced isomerization of retinal in vision, and so forth. Recently, conformationally specific photodissociation dynamics has been discovered in some systems, such as 1iodopropane cation (1-C 3 H 7 I + ) and propanal cation (C3H6O+).1−5 The simplest example so far reported is photodissociation of cis- and trans-formic acid in an argon matrix.2 The CO/CO2 ratio, that is, the branching ratio between CO + H2O and CO2 + H2 channels, varied dramatically depending on the initial conformation. The photolysis of trans-HCOOH, the most stable conformer, at 193 nm predominantly followed the CO + H2O channel with the CO/CO2 ratio of 5.0. The 193 nm photolysis of cisHCOOH, in contrast, gave the CO/CO2 ratio of 0.42. This observation indicated that photodissociation of HCOOH exhibits conformational memory and the initial conformation is maintained during the photodissociation dynamics. However, theoretical results have not been consistent with this observation. A barrier of cis−trans isomerization on the first excited singlet state (S1) PES is only a few kilojoules per mole, which is much lower than the lowest escaping barrier from the potential well of HCOOH on the S1 PES as well as the minimum energy conical intersection (MECI) point for the radiationless decay.6 Moreover, both RRKM and trajectory simulations on the S0 PES indicated that the cis−trans isomerization barrier on the singlet ground state (S0) PES is also too small to maintain the initial conformation until dissociations take place.7 A slight change in the CO/CO2 ratio © 2012 American Chemical Society
Received: June 4, 2012 Accepted: July 3, 2012 Published: July 3, 2012 1900
dx.doi.org/10.1021/jz300728q | J. Phys. Chem. Lett. 2012, 3, 1900−1907
The Journal of Physical Chemistry Letters
Letter
Figure 1. List of stationary structures for the S0, S1, and T1 PESs obtained by the automated search. All structures were optimized at the (MS)CAS(10e,8o)-PT2/aug-cc-pVDZ level. (See the text for details.) In labels of TSs, structures and dissociation channels (DC1−5) connected by each TS are denoted as [X↔Y]. In labels of MSXs, spin−orbit coupling values between two crossing states are also shown. Energy values are relative to the ground state trans-HCOOH optimized at the same level. The structures newly located in this study are marked by †. Cartesian coordinates of all optimized geometries are listed in the Supporting Information. S0/S1-MECI5−7 indicated by * was optimized at the MS-CASPT2 level for the three lowest singlet states because of instability in the two state calculations. S1/S2-MECI1 and 2 indicated by ** were optimized at the MS-CASPT2 level for the four lowest singlet states.
HCOOH. These methods can be applied to multiple PESs in combination with the seam model function and avoiding model function approaches.24,25 By these methods, critical points, that is, minima (MINs) and transition states (TSs) for the ground S0, excited singlet S1, and lowest triplet (T1) PESs as well as S0/ S1 MECIs and S0/T1 and S1/T1 minima on seam of crossing (MSXs), were systematically explored. In the initial search by the GRRM and AFIR methods, SA2-CAS(8e,6o)-SCF/6-31G
excited-state roaming intrinsic only to NO3, or can it be found in other systems? Thus, systematic exploration of this mechanism is also required. In the present study, two automated reaction path search methods, the global reaction route mapping22 (GRRM) and the artificial force induced reaction23 (AFIR), were applied to the three lowest PESs of HCOOH to shed light on the mystery about conformational memory in photodissociation of 1901
dx.doi.org/10.1021/jz300728q | J. Phys. Chem. Lett. 2012, 3, 1900−1907
The Journal of Physical Chemistry Letters
Letter
Figure 2. Potential energy profiles for important pathways. (See Figure 1 for corresponding structures and computation levels.) Connections on the S0, S1, S2, and T1 PESs are shown in blue, red, yellow, and green lines, respectively. MECIs between the singlet states are highlighted with blue cones, whereas MSXs between different spin states are indicated with yellow crosses.
of some key structures were further calculated by UCCSD and UCCSD(T) methods using the Gaussian 09 programs.29 Figure 1 lists all of obtained structures for the S0, S1, and T1 PESs. On the S0 PES, 7 MINs and 16 TSs were located, where all roaming TSs, that is, S0-TS4, S0-TS5, S0-TS6, S0-TS11, and S0-TS15, were newly obtained. These roaming TSs have very similar energies to those for corresponding radical dissociation channels, and this indicates that the roaming species are radicals. On the S1 PES, 4 MINs and 9 TSs are shown. In previous studies, HO-C-OH (S1-MIN3, S1-MIN4) and TSs 4− 9 were not considered. On the T1 PES, structures related to HO-C-OH (T1-MIN1 and T1-TS1), H2CO2 (T1-MIN4 and T1TS6) and H2O·····CO (T1-MIN5 and T1-TS8) as well as TSs 7, 9, and 10 are new. Concerned with MECIs, only S0/S1-MECIs 2, 8, 10, and 11 were previously known. In addition to S0/S1MECIs, we located two S1/S2-MECIs that may play an essential role in the mechanism of the conformational memory, as discussed below. About MSXs, only S1/T1-MSX2 and S1/T1MSX3 were reported so far. We emphasize here that this structure list was obtained on the basis of the systematic search by using GRRM and AFIR. Hence, this list is expected to cover all important channels. This was found to be true at least for experimentally known photodissociation channels of HCOOH, as demonstrated below. Although the employed computation level is moderate and not necessarily quantitative, it is enough to provide qualitative overview of the potential energy landscape. Each important part suggested by the qualitative landscape was subsequently examined at higher levels such as MRCI and CCSD with zero-point energy (ZPE) corrections for quantitative discussions. This multilevel treatment worked very well, as shown below. We were able to choose candidates for important thresholds, and further examinations of them gave reasonable assignments for all of the thresholds with nice
and CAS(8e,6o)-SCF/6-31G levels were employed for the singlet states and the triplet state, respectively, where SA2 stands for state-average of two states S0 and S1. During these searches, CASSCF-MOs were recalculated with a larger (10e,8o) active space and SA3 and SA2 treatments for the singlet and triplet states, respectively, to find better CASSCF solutions, as described in a previous paper.24 On the S0 PES, the GRRM search was applied only to formic acid (HCOOH) and dihydroxylcarbene (HO-C-OH) species. In other words, pathways connecting very high-energy peroxide species were not considered. On the other PESs, including conical intersection and seam of crossing hypersurfaces, the GRRM search was applied only to HCOOH species. In addition, associative (A + B → X (+ Y) type) pathways between two fragments; that is, reverse reactions of photodissociation channels were explored systematically by the AFIR method. Unless otherwise noted, all obtained structures were finally fully optimized at the MS-CAS(10e,8o)-PT2/aug-cc-pVDZ level for the two singlet states and at the CAS(10e,8o)-PT2/aug-ccpVDZ level for the triplet state, where reference functions were obtained by SA2-CASSCF (with equal weight) and CASSCF calculations, respectively, and excitations from all occupied orbitals except for 1s in carbon and oxgen atoms were considered in PT2. In MECI optimization, the branching plane update method was used to optimize MECIs accurately without derivative coupling vector calculations.26 The shift parameter 0.3 was employed in PT2 calculations to avoid the intruder state problem. Spin−orbit coupling values were computed at the SA3-CAS(10e,8o)-SCF/aug-cc-pVDZ level averaging the S0, S1, and T1 states with equal weight. Potential energies and gradients required in automated searches as well as in optimizations were computed by the Molpro2010 program.27 Using these quantities, all geometry displacements were treated by a developmental version of the GRRM program.28 Energies 1902
dx.doi.org/10.1021/jz300728q | J. Phys. Chem. Lett. 2012, 3, 1900−1907
The Journal of Physical Chemistry Letters
Letter
Table 1. Estimates of Relative Energy Values at Important Structures (in kJ/mol, Including ZPE below) on the S1 and T1 PESs a
CASPT2/aug-cc-pVDZ CASPT2/aug-cc-pVTZb MRCISD(Q)/aug-cc-pVTZb UCCSD/aug-cc-pVTZ UCCSD(T)/aug-cc-pVTZc ZPEd
S0-MIN1
S1-MIN1
S1-TS2
S1-TS3
T1-TS3
T1-TS4
T1-TS7
S0/T1-MSX5e
0.0 0.0 0.0 0.0 0.0 88.9
429.4 429.8 442.0
474.6 476.3 498.6
521.5 518.7 541.4
79.6
72.5
57.4
459.6 473.3 477.5 475.9 475.0 68.2
459.6 463.4 485.6 480.6 479.5 59.4
452.5 454.2 477.5 473.7 472.4 60.0
470.3/470.3 476.2/473.1 482.4/483.5 460.0/460.0 459.9/464.4 73.0f
a
Active space includes 10 electrons and 8 orbitals. bActive space includes 12 electrons and 10 orbitals. Single-point calculation at the CAS(10e,8o)PT2/aug-cc-pVDZ optimized geometry. cSingle-point calculation at the UCCSD optimized geometry. dZPE at the CAS(10e,8o)-PT2/aug-cc-pVDZ level. eS0 energy/T1 energy. fZPE at S0/T1-MSX5 was estimated by diagonalizing the projected average Hessian matrix (which is equivalent to Hessians for two crossing states within the first-order intersection space)33 in which projections of three translational vectors, three rotational vectors, and the difference gradient vector are eliminated. Therefore, ZPE here is the same for S0 and T1.
virtually zero at ∼252 nm. At the present level of theory, as seen in Figure 2, the excitation energy required in the 0−0 transition for trans-HCOOH is 438.7 kJ/mol. If the ZPE (Table 1) is considered, then it is reduced to 429.4 kJ/mol. Although this value is smaller than that corresponding to the experimental origin band at ∼267 nm (448 kJ/mol), a higher MRSDCI(Q) level calculation gave a reasonable estimate 442.0 kJ/mol, as shown in Table 1. Direct OH Dissociation Channel on the S1 PES at High Energy. Figure 2 shows that S1-TS2 for the direct OH dissociation can be reached at 491 kJ/mol, above which the system in the S1 state will dissociate very efficiently to produce OH (DC4) with high quantum yield.34 As discussed below, this TS is not related to the 252 nm (475 kJ/mol) threshold. We will discuss this channel in more detail in connection to a pathway via triplet in a subsequent paragraph. Above 532 kJ/ mol (an experimental threshold),35 a direct O−H bond dissociation channel (DC5) opens: S1 → S1-TS3 → HCOO + H. Dissociation Channels below 252 nm; Dissociation via the T1 PES. Now we look for a channel or channels for dissociation below the S1-TS2 threshold. As seen in Figure 2, in the energy range just below S1-TS2, there is no accessible critical point on the S1 PES except for S1-MIN1, S1-MIN2, and S1-TS1. Hence, the molecule with energy below the S1-TS2 is expected to stay on the S1 PES for a long time in the area of trans- and cis-HCOOH structure. The S1 PES and the T1 PES are very close in energy because both of these two are the n→ π* state. Such a long time oscillation around the bottom of these HCOOH minima on the S1 PES should earn enough probability of intersystem crossing (ISC) to the T1 PES. Although spin−orbit coupling between these two states is very small (
