Top Banner
This journal is c the Owner Societies 2011 Phys. Chem. Chem. Phys., 2011, 13, 6145–6155 6145 Cite this: Phys. Chem. Chem. Phys., 2011, 13, 6145–6155 Photodynamical simulations of cytosine: characterization of the ultrafast bi-exponential UV deactivationw Mario Barbatti,* ab Ade´lia J. A. Aquino, ac Jaroslaw J. Szymczak,z a Dana Nachtigallova´ d and Hans Lischka* a Received 27th July 2010, Accepted 27th January 2011 DOI: 10.1039/c0cp01327g Deactivation of UV-excited cytosine is investigated by non-adiabatic dynamics simulations, optimization of conical intersections, and determination of reaction paths. Quantum chemical calculations are performed up to the MR-CISD level. Dynamics simulations were performed at multiconfigurational level with the surface hopping method including four electronic states. The results show the activation of four distinct reaction pathways at two different subpicosecond time scales and involving three different conical intersections. Most trajectories relax to a minimum of the S 1 state and deactivate with a time constant of 0.69 ps mainly through a semi-planar conical intersection along the n O p* surface. A minor fraction deactivate along pp* regions of the S 1 surface. Sixteen percent of trajectories do not relax to the minimum and deactivate with a time constant of only 13 fs. 1. Introduction Upon UV excitation, the five naturally occurring nucleobases return to the ground state by internal conversion at an ultrafast time scale ranging from half a picosecond to few picoseconds. 1–5 In general, ultrafast decay depends on the existence of reaction pathways connecting the Franck–Condon region to the seam of conical intersections between the excited and ground states where radiationless processes can occur. The characterization of these pathways has led to a large amount of theoretical work not only for the five nucleo- bases, 1–14 but also for their isomers, 6,15 substituted species, 7,16,17 and base models. 18,19 Significant progress has been achieved with photodynamical simulations, 20–22 which describe the excited-state time evolution and the most accessed reaction pathways explicitly. Excited-state dynamics simulations are still a major challenge in computational chemistry requiring a proper description of multiple electronic excited states and their non-adiabatic couplings. At the same time, they should keep computational costs under strict control as to allow dynamics propagation for thousands of femtoseconds. Despite the difficulties, ab initio 21–25 and semiempirical 20,26–31 non-adiabatic dynamics simulations have been recently reported for all nucleobases. In the gas phase, the excited-state lifetime of cytosine measured by different groups present somewhat divergent results. Kang et al. 32 (pump: 267 nm; probe: 800 nm) reports a single time constant decay of 3.2 ps. Canuel et al. 33 (pump: 267 nm; probe: 2 400 nm) reports two time constants, 0.16 ps and 1.86 ps. Ullrich et al. 34 (pump: 250 nm; probe: 200 nm) distinguishes three time constants, a very fast decay occurring in less than 0.05 ps, another component of 0.82 ps and a third component of 3.2 ps. Recently, Kosma and co-workers have shown that the excited-state lifetime strongly depends on the excitation wavelength, varying from 3.8 ps to 1.1 ps for pump wavelengths spanning the range from 260 to 290 nm. 35 In common, all these sets of results indicate that cytosine relaxation takes place within one to three picoseconds after the excitation. Additionally, Ullrich et al. 34 have measured the time dependent photoelectron spectra of cytosine and other nucleobases in gas phase. The comparison between the pyrimidine bases cytosine, uracil and thymine clearly indicates that cytosine deactivates in a distinct way producing more energetic photoelectrons. As typical for species deactivating by internal conversion, cytosine fluorescence quantum yield is very small in neutral a Institute for Theoretical Chemistry, University of Vienna, Waehringerstrasse 17, A 1090 Vienna, Austria. E-mail: Mario [email protected], Hans [email protected] b Max-Planck-Institut fu ¨r Kohlenforschung, Kaiser-Wilhelm-Platz 1, D-45470 Mu ¨lheim an der Ruhr, Germany c Institute of Soil Research, University of Natural Resources and Applied Life Sciences Vienna, Peter-Jordan-Straße 82, A-1190 Vienna, Austria d Institute of Organic Chemistry and Biochemistry, Academy of Sciences of the Czech Republic, Flemingovo nam. 2, CZ-16610 Prague 6, Czech Republic w Electronic supplementary information (ESI) available: Experimental time constants, conical intersection notations, molecular orbitals, geometries and pathways at CASSCF level, Cartesian coordinates. See DOI: 10.1039/c0cp01327g z Present address: Department of Chemistry, University of Basel, Klingelbergstrasse 80, 4056 Basel, Switzerland. PCCP Dynamic Article Links www.rsc.org/pccp PAPER Downloaded by Texas Technical University on 04 May 2011 Published on 24 February 2011 on http://pubs.rsc.org | doi:10.1039/C0CP01327G View Online
11

Citethis:Phys. Chem. Chem. Phys.,2011,13 ,61456155 …homepage.univie.ac.at/hans.lischka/pub_domain/Cytosine_dyn_PCCP_… · This ournal is c the Owner Societies 2011 Phys. Chem.

Jun 08, 2018

Download

Documents

hoangthuy
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: Citethis:Phys. Chem. Chem. Phys.,2011,13 ,61456155 …homepage.univie.ac.at/hans.lischka/pub_domain/Cytosine_dyn_PCCP_… · This ournal is c the Owner Societies 2011 Phys. Chem.

This journal is c the Owner Societies 2011 Phys. Chem. Chem. Phys., 2011, 13, 6145–6155 6145

Cite this: Phys. Chem. Chem. Phys., 2011, 13, 6145–6155

Photodynamical simulations of cytosine: characterization of the ultrafast

bi-exponential UV deactivationw

Mario Barbatti,*ab

Adelia J. A. Aquino,ac

Jaroslaw J. Szymczak,zaDana Nachtigallova

dand Hans Lischka*

a

Received 27th July 2010, Accepted 27th January 2011

DOI: 10.1039/c0cp01327g

Deactivation of UV-excited cytosine is investigated by non-adiabatic dynamics simulations,

optimization of conical intersections, and determination of reaction paths. Quantum chemical

calculations are performed up to the MR-CISD level. Dynamics simulations were performed at

multiconfigurational level with the surface hopping method including four electronic states.

The results show the activation of four distinct reaction pathways at two different subpicosecond

time scales and involving three different conical intersections. Most trajectories relax to a

minimum of the S1 state and deactivate with a time constant of 0.69 ps mainly through a

semi-planar conical intersection along the nOp* surface. A minor fraction deactivate along pp*regions of the S1 surface. Sixteen percent of trajectories do not relax to the minimum and

deactivate with a time constant of only 13 fs.

1. Introduction

Upon UV excitation, the five naturally occurring nucleobases

return to the ground state by internal conversion at an

ultrafast time scale ranging from half a picosecond to few

picoseconds.1–5 In general, ultrafast decay depends on the

existence of reaction pathways connecting the Franck–Condon

region to the seam of conical intersections between the excited

and ground states where radiationless processes can occur.

The characterization of these pathways has led to a large

amount of theoretical work not only for the five nucleo-

bases,1–14 but also for their isomers,6,15 substituted species,7,16,17

and base models.18,19 Significant progress has been achieved

with photodynamical simulations,20–22 which describe the

excited-state time evolution and the most accessed reaction

pathways explicitly.

Excited-state dynamics simulations are still a major

challenge in computational chemistry requiring a proper

description of multiple electronic excited states and their

non-adiabatic couplings. At the same time, they should keep

computational costs under strict control as to allow dynamics

propagation for thousands of femtoseconds. Despite the

difficulties, ab initio21–25 and semiempirical20,26–31 non-adiabatic

dynamics simulations have been recently reported for all

nucleobases.

In the gas phase, the excited-state lifetime of cytosine

measured by different groups present somewhat divergent

results. Kang et al.32 (pump: 267 nm; probe: 800 nm) reports

a single time constant decay of 3.2 ps. Canuel et al.33 (pump:

267 nm; probe: 2 � 400 nm) reports two time constants, 0.16 ps

and 1.86 ps. Ullrich et al.34 (pump: 250 nm; probe: 200 nm)

distinguishes three time constants, a very fast decay occurring

in less than 0.05 ps, another component of 0.82 ps and a third

component of 3.2 ps. Recently, Kosma and co-workers have

shown that the excited-state lifetime strongly depends on the

excitation wavelength, varying from 3.8 ps to 1.1 ps for pump

wavelengths spanning the range from 260 to 290 nm.35

In common, all these sets of results indicate that cytosine

relaxation takes place within one to three picoseconds after the

excitation. Additionally, Ullrich et al.34 have measured the

time dependent photoelectron spectra of cytosine and other

nucleobases in gas phase. The comparison between the pyrimidine

bases cytosine, uracil and thymine clearly indicates that

cytosine deactivates in a distinct way producing more energetic

photoelectrons.

As typical for species deactivating by internal conversion,

cytosine fluorescence quantum yield is very small in neutral

a Institute for Theoretical Chemistry, University of Vienna,Waehringerstrasse 17, A 1090 Vienna, Austria.E-mail: Mario [email protected],Hans [email protected]

bMax-Planck-Institut fur Kohlenforschung,Kaiser-Wilhelm-Platz 1, D-45470 Mulheim an der Ruhr, Germany

c Institute of Soil Research, University of Natural Resources andApplied Life Sciences Vienna, Peter-Jordan-Straße 82,A-1190 Vienna, Austria

d Institute of Organic Chemistry and Biochemistry,Academy of Sciences of the Czech Republic, Flemingovo nam. 2,CZ-16610 Prague 6, Czech Republic

w Electronic supplementary information (ESI) available: Experimentaltime constants, conical intersection notations, molecular orbitals,geometries and pathways at CASSCF level, Cartesian coordinates.See DOI: 10.1039/c0cp01327gz Present address: Department of Chemistry, University of Basel,Klingelbergstrasse 80, 4056 Basel, Switzerland.

PCCP Dynamic Article Links

www.rsc.org/pccp PAPER

Dow

nloa

ded

by T

exas

Tec

hnic

al U

nive

rsity

on

04 M

ay 2

011

Publ

ishe

d on

24

Febr

uary

201

1 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/C

0CP0

1327

GView Online

Page 2: Citethis:Phys. Chem. Chem. Phys.,2011,13 ,61456155 …homepage.univie.ac.at/hans.lischka/pub_domain/Cytosine_dyn_PCCP_… · This ournal is c the Owner Societies 2011 Phys. Chem.

6146 Phys. Chem. Chem. Phys., 2011, 13, 6145–6155 This journal is c the Owner Societies 2011

aqueous solution.36 The quantum yield, however, tends to

increase with the pH and even phosphorescence was observed

in high pH conditions.37 Consistent with the quantum yields

derived from static spectra, time-dependent spectra of cytosine

in water show a time constant of about 1 ps in neutral

solution.38 The time constant is shortened to 0.63 ps in

pH 0.08 and elongated to 13.3 ps in pH 13. (A survey of

experimental time constants for cytosine and cytosine

derivatives under different conditions is given in Table S1 of

the Supplementary Information.)

Also from a theoretical perspective, excited state excitation

and deactivation of cytosine has been subject to several studies

employing different methodologies. There are three well

described S1/S0 conical intersections in cytosine, and the paths

connecting them to the Franck–Condon region and to excited

state minima have been investigated in detail at several

theoretical levels.10,11,16,39–44 A recent review on this topic

can be found in ref. 39.

Two of the three known conical intersections in cytosine were

first characterized by Ismail and co-workers.11 The first conical

intersection has a semi-planar structure with sp3 hybridization of

C6, shortening of the C2–N3 bond and stretching of the C5–C6

bond relative to the ground state minimum (see numbering in

Fig. 1). It occurs in a region of mixing of the nOp*, pp* and

closed shell states, which ends in a triply degeneracy as shown in

ref. 45–47. Throughout this paper we will refer to this conical

intersection as the semi-planar conical intersection. (Since no

generally accepted scheme for labeling conical intersections has

been proposed so far, we give a cross-referenced summary of

notations for conical intersections in cytosine in Table S2 of the

Supplementary Information for the readers’ convenience.)

The second S1/S0 conical intersection is characterized by

puckering of atom N3, which induces a strong out-of-plane

deformation of the amino group. It has been often assigned as

connected to the nNp* state, but it will be shown in this work

that in fact it is reached by a reaction path of pNp* character,

which is stabilized by a twisting of the N3–C4 bond. We will

refer to this structure as the oop-NH2 conical intersection.

The third conical intersection occurs also with involvement of

a pp* state and is characterized by a strong puckering of the C6

atom induced by a twisting around the C5–C6 bond. We refer

to this structure as the C6-puckered conical intersection. This

conical intersection was found by Sobolewski and Domcke48

when investigating guanine-cytosine pairs and its electronic

structure has been investigated in detail by Zgierski et al.10,40

Because of the barrierless diabatic connection to the ground

state, Merchan, Serrano-Andres43,44 and co-authors have

predicted that cytosine should relax to the ground state

directly through the C6-puckered conical intersection. This

prediction is supported by surface hopping dynamics

simulations performed with the OM2 semi-empirical level.30

Nevertheless, dynamics simulations with multiple spawning24

at CASSCF(2,2) and surface hopping31 at AM1/CI(2,2) levels

have found only small fractions of deactivation at the

C6-puckered conical intersection and a major part of deactivation

at the oop-NH2 conical intersection.

In addition to these three conical intersections, we have

optimized a new S1/S0 conical intersection in cytosine related

to a pOpO* state. It is characterized by a strong out-of-plane

deformation of the O atom analogous to uracil12 and

thymine.9 However, because of its relatively high energy, it is

not expected to play any relevant role in deactivation of

low-energy excited cytosine. It will be referred to as the

oop-O conical intersection.

This large set of theoretical investigations of cytosine have

demonstrated the existence of multiple pathways for internal

conversion available either by direct relaxation to the conical

intersections or by a stepwise process first relaxing into the S1minimum and then moving to the intersection seam. In the

present work the photodeactivation of UV-excited cytosine

has been addressed by non-adiabatic dynamics simulations

with an electronic structure level significantly superior to those

employed in previous simulations, with a detailed control of

the initial conditions as to simulate as closely as possible the

experimental settings, and with extensive comparisons to other

theoretical and experimental results in the literature. Moreover,

optimization of stationary points, conical intersections and

determination of reaction pathways has been performed

up to multireference configuration interaction with singles

and doubles (MR-CISD) level. The results allow drawing a

comprehensive scenario for deactivation of cytosine based on a

bi-exponential sub-picosecond decay involving four distinct

reaction pathways and three different regions of the seam of

conical intersections. The results also show two singular features

in the deactivation process of cytosine: first a fraction of cytosine

population returns to the ground state within 10 fs, one of the

fastest internal conversion among organic molecules; second, the

deactivation at a conical intersection involving three states.

2. Computational details

Mixed quantum-classical dynamics simulations49 were performed

for cytosine at the complete active space self-consistent field

(CASSCF) level. The active space was composed of fourteen

electrons in ten orbitals [CASSCF(14,10)]. At the ground state

Fig. 1 Structures of the ground and S1 minima of cytosine with

selected geometrical parameters optimized at CASSCF(14,10)/6-31G*

and MR-CISD(6,5)/CASSCF(14,10)/6-31G* (in parentheses) levels of

theory. Valence-bond structures based on bond distances. Bond

distances are given in A.

Dow

nloa

ded

by T

exas

Tec

hnic

al U

nive

rsity

on

04 M

ay 2

011

Publ

ishe

d on

24

Febr

uary

201

1 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/C

0CP0

1327

GView Online

Page 3: Citethis:Phys. Chem. Chem. Phys.,2011,13 ,61456155 …homepage.univie.ac.at/hans.lischka/pub_domain/Cytosine_dyn_PCCP_… · This ournal is c the Owner Societies 2011 Phys. Chem.

This journal is c the Owner Societies 2011 Phys. Chem. Chem. Phys., 2011, 13, 6145–6155 6147

minimum geometry, these orbitals are three n, four p,and three p* orbitals (see Fig. S1 of the Supplementary

Information). State averaging was performed over four states

(SA-4) and the 6-31G* basis set50 was employed. Analytical

energy gradients and non-adiabatic coupling vectors were

computed by the procedures described in ref. 51–55.

The classical equations were integrated with the Velocity-

Verlet algorithm56 using a 0.5 fs time step. The quantum

equations were integrated with the 5th-order Butcher

algorithm57 with a 0.01 fs time step. The partial coupling

approximation58 was used to reduce the number of non-

adiabatic couplings to be computed every time step. Decoherence

effects were introduced by the model discussed in ref. 59

(a = 0.1 hartree). Non-adiabatic events were taken into

account by the surface hopping fewest-switches algorithm.60,61

Results were analyzed in terms of the Cremer-Pople (CP)

parameters62 using the Boeyen’s conformer classification

scheme.63 The CP parameters quantify the degree (Q) and

the kind (y and f) of ring puckering. In particular, Q = 0 A

indicates planar structures and y and f parameters can be used

to classify the ring conformation as boat (B), envelope (E),

chair (C), half-chair (H), screw-boat (S), and twisted-boat (T).

Initial geometries and velocities were generated by a Wigner

distribution treating each nuclear coordinate as a harmonic

oscillator in the ground state. Harmonic frequencies computed

at the CASSCF(14,10)/6-31G* level were scaled by a factor

l = 0.90452, which was derived by comparison to experimental

frequencies given in ref. 64. One thousand random geometries

and velocities were generated. Single point calculations

were performed for each of them and used to compute the

absorption cross section with the semi-classical approximation

described in ref. 65. Three excited states were included and

Lorentzian line shapes with width 0.1 eV were employed. The

simulated spectrum (see Fig. 2) shows a single band dominated

by the first pp* transition and centered at 5.49 eV.

To perform the simulations as closely as possible to the

experimental conditions, it should be taken into account

that time-resolved spectroscopic experiments33,66 have been

performed by pumping cytosine at 267 nm near the center of

the first absorption band located at 260 nm.67 Thus, in the

construction of initial conditions excitation energies have been

restricted to the range 5.25 � 0.25 eV, near the center of the

simulated spectrum. Besides this energy restriction, initial

conditions were randomly selected in terms of transition

probabilities into the three excited states. From the original

3000 possible initial conditions (1000 random points starting

in each of the three excited states), this procedure selected

172 initial conditions, 49 starting in S1, 105 starting in S2, and

18 starting in S3. The number of trajectories finally

computed—30 starting in S1, 64 starting in S2 and 11 starting

in S3—was chosen to keep this proportion. The 105

trajectories were propagated for a maximum time of 1.2 ps.

Trajectories that returned to the ground state and stayed there

for more than 50 fs were terminated.

The relaxation of the initial dynamics was also simulated at

the resolution-of-identity coupled cluster to the second-order

(RI-CC2)68–70 method with the SVP basis set. Initial conditions

were selected from the center of the first band of the absorption

spectrum, within�0.25 eV range around the vertical excitation

into the pp* state (4.88 eV). Fifteen trajectories were computed

for a maximum of 20 fs with time step 0.5 fs.

Minima and conical intersections were optimized at the

CASSCF level described above and also at the MR-CISD

level. The reference space was composed by a complete active

space with six electrons in five orbitals. This space was built

based on the analysis of CASSCF(14,10) calculations for the

minima and conical intersections. Orbitals with natural

occupation lower than 0.1 in all those geometries were moved

to the virtual space and orbitals with natural occupation

higher than 0.9 were moved to the doubly occupied space.

Sixteen orbitals were kept in the frozen space and generalized

interacting space restrictions were adopted.71 TheMR-CISD(6,5)

calculations were performed with the orbitals obtained at

SA-4-CASSCF(14,10)/6-31G* level. Davidson corrections52,72,73

(+Q) were additionally computed for single point calculations.

Cartesian coordinates for all optimized structures are given in

the Supplementary Information.

MRCI and CASSCF calculations were performed with the

COLUMBUS program system74–76 and for the dynamics

simulations the NEWTON-X program77,78 was used. RI-CC2

and DFT calculations were performed with the TURBOMOLE

program.79 Cremer-Pople parameters were computed with

the PLATON program.80

3. Results and discussion

3.1 The potential energy surface of cytosine

3.1.1 Vertical excitation energies. The geometries of the

ground state minimum of cytosine optimized at CASSCF and

MR-CISD levels, both using the 6-31G* basis set, are shown

in Fig. 1. The main differences between the two levels are

found in the N3–C4 and at the C6–N1 bond distances, which

are, respectively, stretched by 0.027 A and shrunk by 0.022 A

when optimized at the higher level.

Vertical excitation energies for the lowest electronic states of

cytosine are given in Table 1. At the CASSCF level employed

in the dynamics calculations as well as in the MR-CISD level,

the np* state is the first singlet excited state and it is closely

followed by the first pp* state. The inclusion of the Davidson

correction (+Q) to the MR-CISD level stabilizes the pp*state, which becomes the first excited state. This result is also

confirmed at the CC2 level. We shall discuss later in this

section that this inversion does not affects the dynamics

results. As usual, the vertical excitation energy of the pp*ionic state is overestimated.81 The effect of this overestimation

on the dynamics results are discussed in the section

‘‘Photophysics of cytosine.’’

The S4 state at the CC2/aug-TZVP level (not shown in

Table 1) is the p3s state. Its transition energy is 5.44 eV and its

oscillator strength is 0.01. Since this state lies 0.8 eV above the

vertical excitation into the pp* state, we do not expect that it

will be involved in the photodynamics of cytosine starting in

the first absorption band.

The simulated spectrum at the CASSCF level is shown in

Fig. 2. The band has a Gaussian profile with amplitude

0.07 A2.molecule�1 and width 1.15 eV, which is in good

agreement with CC2 results65 (0.08 A2.molecule�1, 1.05 eV).

Dow

nloa

ded

by T

exas

Tec

hnic

al U

nive

rsity

on

04 M

ay 2

011

Publ

ishe

d on

24

Febr

uary

201

1 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/C

0CP0

1327

GView Online

Page 4: Citethis:Phys. Chem. Chem. Phys.,2011,13 ,61456155 …homepage.univie.ac.at/hans.lischka/pub_domain/Cytosine_dyn_PCCP_… · This ournal is c the Owner Societies 2011 Phys. Chem.

6148 Phys. Chem. Chem. Phys., 2011, 13, 6145–6155 This journal is c the Owner Societies 2011

The np* states are completely hidden by the pp* intense

absorption. The dashed area indicates the range from where

initial conditions for dynamics were selected, just below the

absorption maximum as in the time-resolved experiments.

While in the ground state minimum the pp* state is slightly

above the np* state at the CASSCF level, in the distorted

geometries sampled by the Wigner distribution, the pp* may

be also found below the np* state. Trajectory simulations

starting in either situation show essentially the same results,

which means that the inverted order of the np* and pp* states

at CASSCF level is not important for the dynamics results.

3.1.2 Excited-state minimum and conical intersections. The

geometry of the S1 excited state minimum optimized at

CASSCF and MR-CISD levels is shown in Fig. 1. Energies

of the ground and excited states are given in Table 2 for these

geometries. In both CASSCF andMRCI levels, the S1 minimum

has nOp* character. Although the ring is still planar, hydrogen

atoms connected to N1 and C6 atoms are displaced out of the

plane. Molecular orbitals for both ground and excited state

minima are shown in Fig. S1 of the Supplementary Information.

It was not possible to optimize the previously reported pp* S1minimum44 at the present CASSCF level, which seems to be

coincident with the S2/S1 minimum on the crossing seam.

Four different minima on the crossing seam (MXSs) have

been optimized at CASSCF and MR-CISD levels. Their

geometries are show in Fig. 3 at MR-CISD level and selected

geometric parameters are given in Table 3. Molecular orbitals

and geometries optimized at CASSCF level are shown in

Fig. S2 and S3 of the Supplementary Information. The lowest

MXS at the CASSCF level (oop-NH2) is characterized by a

strong deformation of the NH2 group out of the ring plane. It

has a screw-boat conformation with puckering of atoms

N3 and C4 (3S4). This conical intersection has been often

characterized as a crossing between the nNp* and closed shell

(cs) states.11,16,24,39,42,44 A closer inspection of the electronic

configurations involved in this conical intersection, however,

indicates that it can be better characterized as a pN3pC4*/cscrossing.46 This can be seen from Fig. 4, where the relevant nN,

p and p* orbitals are displayed for the ground state minimum

and conical intersection geometries. By twisting the N3–C4

bond in moving from the ground-state structure to the

oop-NH2 MXS the lone pair on N3 is rotated out from the

molecular plane, but otherwise keeps its characteristics. It

remains doubly occupied while the single occupied orbital is

indeed the p orbital with a large density at N3. This distinction

is important because it clarifies the origin of this conical

intersection, which corresponds to an ethylenic conical inter-

section arising from the twisting around the N3–C4 bond.

The next MXS in the energetic order of the CASSCF level is

the semi-planar MXS. It is characterized by strong stretching

of the C5–C6 bond accompanied by a strong shrinking of the

C2–N3 bond. A certain degree of pyramidalization is also

observed at the C6 atom. Although the minimum energy

geometry has a twist-boat conformation slightly puckered

(Q = 0.25 A) at atoms N1 and N3 (1T3), we anticipate that

the dynamics simulations will reveal that this section of the

crossing seam spreads over a large region of the conformation

space. Electronically, this MXS is formed by a crossing

between the nOp* and the pp* states, but showing a large

mixing with the closed shell configuration. Indeed, as it has

been discussed in ref. 45 and 47, the S0, S1 and S2 states are

very close to each other for geometries in this region of the

crossing seam space, even forming three state conical

intersections.

The third conical intersection (C6-puckered) is characterized

by a twist around the C5-C6 bond, which leads to a C6

puckering (screw-boat 6S1). This is also an ethylenic conical

intersection with crossing between pC5pC6* and closed shell

states.

The highest energy conical intersection optimized in this

work (oop-O) is characterized by a strong out-of-plane

displacement of the O atom keeping the ring planar or

semi-planar. It is energetically too high in energy to play any

role in the low energy excitation dynamics of cytosine. Besides

Fig. 2 Absorption spectrum into the first band simulated at

CASSCF(14,10)/6-31G* level. Crosses show the oscillator strength

of the vertically excited S1 (nNp*), S2 (pp*) and S3 (nOp*) states. Thedashed area indicates the spectral region where initial conditions were

selected.

Table 1 Theoretical data for the first three vertical excitation levels of cytosine computed with different methods

S1 S2 S3

DE (eV) f DE (eV) f DE (eV) f

CASSCF 5.41 np* 0.002 5.56 pp* 0.085 5.74 np* 0.003MR-CISD 5.72 np* 0.003 5.86 pp* 0.140 7.48 np* 0.019MR-CISD+Q 5.39 pp* 0.003 5.55 np* 0.140 6.80 np* 0.019CC2/SVP 4.88 pp* 0.055 5.04 np* 0.002 5.41 np* 0.001CC2/TZVP 4.71 pp* 0.051 4.95 np* 0.008 5.35 np* 0.009CC2/aug-TZVP 4.61 pp* 0.052 4.87 np* 0.007 5.27 np* 0.010

Dow

nloa

ded

by T

exas

Tec

hnic

al U

nive

rsity

on

04 M

ay 2

011

Publ

ishe

d on

24

Febr

uary

201

1 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/C

0CP0

1327

GView Online

Page 5: Citethis:Phys. Chem. Chem. Phys.,2011,13 ,61456155 …homepage.univie.ac.at/hans.lischka/pub_domain/Cytosine_dyn_PCCP_… · This ournal is c the Owner Societies 2011 Phys. Chem.

This journal is c the Owner Societies 2011 Phys. Chem. Chem. Phys., 2011, 13, 6145–6155 6149

that, cytosine should present other S0/S1 crossings for NH

stretching and CN ring opening. The intersection along the

NH stretching should occur by a crossing between the p3s andthe closed shell states.82 Because of its high vertical excitation

energy, the p3s state in cytosine is not expected to play a

relevant role in the photodynamics following excitation into

the first pp* state. In the case of the ring-opening conical

intersections, their activation usually involves the overcoming

of high energy barriers.83 Dynamics simulations will also show

that they also do not play any role in the deactivation of

cytosine excited into the first absorption band.

The geometries of the MXSs optimized at the MR-CISD

level are rather similar to the ones optimized at the CASSCF

level (Table 3). The differences are, however, large enough

to change the conformation classification. Thus, while the

oop-NH2 MXS has a 3S4 conformation at the CASSCF level,

it has an envelope 3E at the MR-CISD level, which implies a

smaller degree of C4 puckering. Similarly, the puckering of the

N1 atom in C6-puckered MXS decreases when it is optimized

at the MR-CISD level. Because of this, the conformation

changes from 6S1 to6E.

The energetic order of the MXSs also changes at the

MR-CISD level, although the state character remains

the same as in CASSCF (Table 2). At MR-CISD + Q level,

the semi-planar MXS is destabilized by about 0.6 eV while

the C6-puckered is stabilized by about 0.3 eV in comparison

to the CASSCF level. Because of this, the energetic order

is altered and the oop-NH2 and C6-puckered MXSs appear

first with very close energies followed by the semi-planar at

higher energy. The implications of this change will be

discussed later. (Note that because the MXSs were optimized

at the MR-CISD level and the energies were computed at

the MR-CISD+Q level, there is a small energy split between

S0 and S1.)

Table 2 Vertical excitation energies (eV), energies at the S1 minimum and at the S1/S0 MXSs for cytosine obtained with CASSCF and MR-CISDmethods. Values with Davidson correction are given in parentheses. cs—closed shell

Geom. Conform. S0 S1 S2 S3

SA-4-CASSCF(14,10)/6-31G*Min S0 Planar 0.00a cs 5.41 nNp* 5.56 pp* 5.74 nOp*Min S1 Planar 2.25 cs 3.73 nOp* 4.14 pp* 6.07 nOp*MXS oop-NH2 (

3S4) 3.84 cs 3.84 pN3pC4* 6.73 nNp* 7.16 pp*MXS Semi-planar (1T3) 4.09 nOp* + cs 4.09 pp* 4.53 cs + nOp* 7.19 nOp*MXS C6-puckered (6S1) 4.34 cs 4.34 pC5pC6* 6.43 pp* 7.17 nNp*MXS oop-O (3S2) 5.99 cs + nOpO* 5.99 pOpO* 7.87 cs + nOp* 9.83 nOp*MR-CISD(6,5)/SA-4-CASSCF(14,10)/6-31G*Min S0 Planar 0.00b (0.00c) cs 5.72 (5.55) nNp* 5.86 (5.39) pp* 7.48 (6.80) nOp*Min S1 Planar 2.04 (1.66) cs 4.40 (4.53) nOp* 5.89 (4.97) pp* 6.63 (6.64) nOp*MXS oop-NH2 (

3E) 4.12 (3.99) cs 4.12 (4.08) pN3pC4* 6.98 (6.89) nNp* 7.46 (7.23) pp*MXS Semi-planar (1T3) 4.85 (4.79) cs + nOp* 4.85 (4.65) pp* 6.75 (5.93) nOp* 8.47 (8.42) nOp*MXS C6-puckered (6E) 4.17 (3.91) cs 4.17 (4.22) pC5pC6* 6.94 (6.87) pp* 8.03 (7.74) nNp*MXS oop-O (B4,1) 5.65 (5.06) cs 5.65 (5.53) pOpO* 8.31 (7.82) ppO* 8.21 (8.61) (pO*)

2

a E0 = �392.671177 au. b E0 = �393.153516 au. c E0 = �393.234586 au.

Fig. 3 Geometry of four minima on the S1/S0 crossing seam of

cytosine optimized at MR-CISD level. Geometries optimized at

CASSCF level are shown in the Supplementary Information.

Table 3 Characterization of the S1/S0 minima on the crossing seamoptimized at CASSCF and MRCI levels in terms of the Cremer-Popleparameters Q (A), y (1), and f (1) and of selected bond distances (A)

oop-NH2 Semi-planar C6-puckered oop-O

Conf. CASSCFa 3S41T3

6S12S3

MRCIb 3E 1T36E B4,1

Q CASSCF 0.50 0.25 0.52 0.09MRCI 0.52 0.44 0.47 0.18

y CASSCF 62 80 119 114MRCI 58 81 120 100

f CASSCF 140 322 146 272MRCI 125 324 132 185

N1–C2 CASSCF 1.377 1.339 1.589 1.358MRCI 1.377 1.348 1.562 1.454

C2–N3 CASSCF 1.418 1.240 1.319 1.369MRCI 1.422 1.243 1.349 1.364

N3–C4 CASSCF 1.410 1.455 1.355 1.280MRCI 1.439 1.455 1.331 1.291

C4–C5 CASSCF 1.470 1.338 1.381 1.485MRCI 1.456 1.347 1.446 1.465

C5–C6 CASSCF 1.361 1.481 1.471 1.337MRCI 1.358 1.466 1.452 1.359

C6–N1 CASSCF 1.406 1.403 1.335 1.408MRCI 1.412 1.449 1.358 1.356

C2–O7 CASSCF 1.190 1.399 1.197 1.632MRCI 1.196 1.379 1.186 1.385

C4–N8 CASSCF 1.397 1.392 1.379 1.351MRCI 1.401 1.413 1.347 1.352

a CASSCF: SA-4-CASSCF(14,10)/6-31G*. b MRCI: MR-CISD(6,5)/

SA-4-CASSCF(14,10)/6-31G*.

Dow

nloa

ded

by T

exas

Tec

hnic

al U

nive

rsity

on

04 M

ay 2

011

Publ

ishe

d on

24

Febr

uary

201

1 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/C

0CP0

1327

GView Online

Page 6: Citethis:Phys. Chem. Chem. Phys.,2011,13 ,61456155 …homepage.univie.ac.at/hans.lischka/pub_domain/Cytosine_dyn_PCCP_… · This ournal is c the Owner Societies 2011 Phys. Chem.

6150 Phys. Chem. Chem. Phys., 2011, 13, 6145–6155 This journal is c the Owner Societies 2011

3.1.3 Reaction paths. Reaction pathways for internal

conversion of cytosine at the three lowest energy conical

intersections have been computed at the CASSCF (Fig. S4

in the Supplementary information) andMR-CISD+Q (Fig. 5)

levels. All these conical intersections are energetically accessible

after excitation into the pp* state. The reaction paths were

computed from the ground state minimum geometry to the

conical intersections (upper graphs in Fig. S4 and Fig. 5) and

from the ground state minimum geometry to the excited state

minimum and from there to the conical intersections (bottom

graphs). In all cases, they were obtained by linear interpolation

of internal coordinates between optimized structures at the

CASSCF and MR-CISD levels.

The pathways computed at both levels of theory present

essentially the same qualitative features, what should lead to

similar dynamical behaviors. One main difference between the

CASSCF and MR-CISD+Q pathways is related to the S2state of the pathway going to the semi-planar conical inter-

section (Fig. S4 and Fig. 5, top-left). At CASSCF level,

the minimum on the crossing seam is much closer to the

three-state crossing region45 than at MR-CISD+Q level, but

since the initial relaxation should follow the pp* state, which

remains in similar positions for both methods, this feature

should not affect the outcome of the dynamics.

The dynamics simulations at CASSCF level discussed in the

next sections will show that a pathway similar to Fig. S4

(top-left) is initially followed by cytosine. At the conical

intersection a small fraction of the trajectories return to the

ground state, but most of them relax to the minimum of S1state and can later convert to the ground state by any of the

three pathways shown in Fig. S4 (bottom). Naturally, the

fraction of trajectories that will follow each pathway will

depend on topographical details of the surfaces, mainly the

height of the energy barriers separating the minimum and

Fig. 4 Electronic configurations of the nN, p and p* orbitals at (a) theground state minimum geometry (S0 state) and at (b) the oop-NH2

MXS (S0 and S1 states).

Fig. 5 Linearly interpolated pathways between the Franck–Condon (FC) region and the three lowest-energy MXSs (top) and between the FC

region, the S1 minimum and the three MXSs (bottom). Computed at MR-CISD(6,5)+Q/SA-4-CASSCF(14,10)/6-31G* level with geometries

optimized at MR-CISD level. Only three states are shown for clarity. The same pathways computed at CASSCF level are shown in the

Supplementary Information.

Dow

nloa

ded

by T

exas

Tec

hnic

al U

nive

rsity

on

04 M

ay 2

011

Publ

ishe

d on

24

Febr

uary

201

1 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/C

0CP0

1327

GView Online

Page 7: Citethis:Phys. Chem. Chem. Phys.,2011,13 ,61456155 …homepage.univie.ac.at/hans.lischka/pub_domain/Cytosine_dyn_PCCP_… · This ournal is c the Owner Societies 2011 Phys. Chem.

This journal is c the Owner Societies 2011 Phys. Chem. Chem. Phys., 2011, 13, 6145–6155 6151

conical intersection. As it has been discussed by Blancafort42

the differences between barriers along CASSCF pathways and

pathways computed at higher levels, in this case CASPT2, are

of about 0.2 eV, which are in the accuracy limit of not only

CASSCF but also of most of the other methods.

3.2 Dynamics results

Fig. 6 shows the time evolution of the occupation of each

adiabatic state. At time zero, cytosine is in the pp* state, whoseintensity is distributed as 29% to S1, 61% to S2 and 10% to S3(see discussion about initial conditions in the ‘‘Computational

details’’ section). Non-adiabatic events take place already in

the first 10 fs. S2, the initially most strongly occupied state,

transfers part of the population to S1 and to S3, reflecting the

exchange of the diabatic character of the three states. S3 is

quickly depopulated within 20 fs. The occupation of the first

excited state reaches a maximum of about 60% in 20 fs and

remains oscillating around this level during the first 100 fs.

After 100 fs, the transfer to the ground state is intensified.

The ground state is populated bi-exponentially. Its

occupation can be fitted with the three-parameter function:

f2(t) = 1 � a exp(�t/t1) � (1 � a)exp(�t/t2),

where t is the time and t1 and t2 are the two time constants.

The amplitude a is the fraction of the initial excited state

population following the pathways with time constant t1,while (1 � a) is the fraction following the pathways with t2.The fitting procedure results in a very fast t1 = 13 fs decay

component followed by a = 16% of the population and a

slower t2 = 0.7 ps component followed by (1 � a) = 84% of

the population (Table 4). This gives an average lifetime of

hti = at1 + (1 � a)t2 = 0.58 ps.

In this analysis, we assumed that trajectories that were still

in the excited state at the end of the simulation (1.2 ps) should

deactivate within the exponential decay t2. An alternative

interpretation is that these trajectories are not part of the

second exponential decay and they will deactivate with a

longer time constant. In this case, the ground state population

can be fitted with the function:

f3(t) = 1 � a1 exp(�t/t1) � a2 exp(�t/t2)� (1 � a1 � a2)exp(�t/t3).

With this procedure, 13% of trajectories decay with the

shortest time constant t1 = 9 fs, 74% decay with

t2 = 0.53 ps and the remaining 13% decay with t3 = 3.1 ps

(Table 4). The average lifetime becomes hti= a1t1 + a2t2 +(1 � a1 � a2)t3 = 0.79 ps.

With exception of oop-O, all other conical intersections

described in the previous sections were origin of S1/S0 hopping

events during the dynamics. oop-NH2 conical intersection was

observed with 3S4 and3H4 conformations. It was responsible

for deactivation of 7% of the trajectories (Table 5).

C6-puckered conical intersection was observed in 8% of the

trajectories. They were associated with puckering at the C6

atom, being formed with 6S1,6E, 6,3B, and 6S5 conformations.

The major fraction of trajectories, 68%, deactivated at the

semi-planar conical intersection, with a large variety of

conformations and puckering degrees. From this amount,

16% occurred within 10 � 2 fs giving origin to the t1 time

constant, while 52% contributed to the t2 time constant. When

the semi-planar intersection occurred with larger degree of

puckering, they were concentrated around the 6E conformation.

The 1T3 conformation, corresponding to the minimum on the

crossing seam of this region, was observed in only few cases.

When the degree of puckering was small, these intersections

were concentrated at the 6T2 and 4T2 conformations. The

large variety of puckering conformations observed in the

C6-puckered conical intersections is a consequence of the fact

that these intersections are mainly formed by in-plane rather

than out-of-plane ring deformations. The remaining 17% of

trajectories did not return to ground state within the 1.2 ps

simulation time. About 20% of S1/S0 hopping events took

place with a degree of puckering smaller than 0.15 A (see Fig. 6

bottom), corresponding to structures with planar or quasi

planar rings. Because of the proximity between S0, S1 and S2in the region of semi-planar conical intersection, 21% of all

trajectories returned to the ground state in a S2 - S0 hopping

event (pp* - cs, with np* between them). This enforces the

importance of a proper multistate treatment of the surface

hopping algorithm in the case of cytosine, which is often

restricted to two-states approximations.

Typically, trajectories quickly relaxed along the pp* state as

illustrated in Fig. 7. The closed shell state is strongly destabilized

during this initial process and after about only 10 fs cytosine is

brought to a region of crossing between pp*, nOp* and cs

states. There, it can be either non-adiabatically transferred

to the S0 state (Fig. 7 top), which happened in 16% of

trajectories, or to remain excited (Fig. 7 bottom) as in the

remaining 84%. Note that a hopping algorithm based on

energy thresholds instead of based on non-adiabatic transition

probabilities would completely fail to describe this process.

Fig. 6 Time evolution of the population of the ground and excited

states of cytosine (top). Degree of ring puckering at the S1/S0 hopping

time (bottom).

Dow

nloa

ded

by T

exas

Tec

hnic

al U

nive

rsity

on

04 M

ay 2

011

Publ

ishe

d on

24

Febr

uary

201

1 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/C

0CP0

1327

GView Online

Page 8: Citethis:Phys. Chem. Chem. Phys.,2011,13 ,61456155 …homepage.univie.ac.at/hans.lischka/pub_domain/Cytosine_dyn_PCCP_… · This ournal is c the Owner Societies 2011 Phys. Chem.

6152 Phys. Chem. Chem. Phys., 2011, 13, 6145–6155 This journal is c the Owner Societies 2011

The trajectories that remained in the excited state after this

first encounter with S0 were transferred to the nOp* state

within the next 100 fs. They relaxed to the energy minimum

of this state and remained there for an average time of about

0.6 ps before being finally deactivated. During this time,

small energy gaps to the ground state often occurred. The

non-adiabatic transition probability is small enough to

guarantee that multiple of such encounters occur without

returning to the ground state. The reason for this rests on

the shape of the potential energy surface in the region between

the S1 minimum and the semi-planar conical intersection. The

sloped conical intersection is reached in an up-hill motion,

which decreases the efficiency of the non-adiabatic transitions.

Because of this, other pathways to the ground state through

pp*/cs conical intersections (oop-NH2 and C6-puckered

conical intersections) start to compete with the nOp*/cspathway. As discussed in section ‘‘Reaction paths,’’ the

activation of the pp*/cs pathways starting from the nOp*minimum depends on overcoming energetic barriers to the

biradical pp* states. When this is achieved, the conical inter-

section to the ground state can be easily accessed because of its

peaked shape and the internal conversion occurs quickly,

usually in less than 100 fs after the barrier is overcome.

The occurrence of a conical intersection with the ground

state as soon as 10 fs after the photoexcitation is an unexpected

feature for a large molecule like cytosine. It is certainly one of

the fastest internal conversion processes among organic

molecules, even faster than the exceptionally fast decay of

ethylene whose lifetime is about 38 fs.84 To check whether this

was not an artifact of the CASSCF method induced by

artificially large energy gradients in the Franck–Condon

region, we have simulated the initial relaxation of cytosine

with dynamics simulations performed with the RI-CC2/SVP

method. Exactly as predicted at CASSCF level, all CC2

trajectories moved into the semi-planar S1/S0 crossing within

about 10 fs. The average potential energy over all trajectories

plotted as a function of time is shown in Fig. 8. This figure also

shows the shortening of the C2-N3 bond length in the beginning

of the dynamics typical for the semi-planar conical intersection.

3.3 Photophysics of cytosine

Fig. 9 schematically illustrates our main findings concerning

the photodynamics of cytosine excited into the first singlet pp*

Table 4 Time constants for cytosine relaxation after UV excitation in gas phase. SH—surface hopping; MS—multiple spawning. Results fromref. 35 are those for pump wavelength 280 nm, which has the closest correspondence to the spectral region excited in the present work

t1 (ps) t2 (ps) t3 (ps)

ExperimentalRef. 32 — 3.2Ref. 34 o0.05 0.82 3.2Ref. 33 0.16 � 0.02 1.86 � 0.19Ref. 35 o0.1 1.2 55TheoreticalSH/OM2 ref. 30 0.04 0.37MS/CAS(2,2) ref. 24 o0.02 B0.8SH/CAS(14,10), present worka 0.013 � 0.001 0.688 � 0.002SH/CAS(14,10), present workb 0.009 � 0.001 0.527 � 0.005 3.08 � 0.04

a Bi-exponential fitting: a = 0.16, (1 � a) = 0.84. b Tri-exponential fitting: a1 = 0.13, a2 = 0.74, (1 � a1 � a2) = 0.13.

Table 5 Main conical intersections accessed for S1/S0 deactivation. Conical intersection energies computed at same level as the dynamicssimulations are given in parentheses. SH—surface hopping; MS—multiple spawning; nr—not reported

Semi-planar (early) Semi-planar (late) oop-NH2 C6-puckered

SH/OM2a 0% (4.34) 0% (4.34) 0% (nr) 100% (3.64)MS/CAS(2,2)b 15% (5.25) 0% (5.25) 65% (4.35) 5% (3.98)SH/CAS(14,10)c 16% (4.09) 52% (4.09) 7% (3.84) 8% (4.34)

a Ref. 30. b Ref. 24. c Present work.

Fig. 7 Time evolution of the potential energy of the ground and

excited states of cytosine in the beginning of the dynamics for two

selected trajectories. The circles indicate the current state in each time

step. Full circles—pp* state; open circles—closed shell state (cs).

Dow

nloa

ded

by T

exas

Tec

hnic

al U

nive

rsity

on

04 M

ay 2

011

Publ

ishe

d on

24

Febr

uary

201

1 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/C

0CP0

1327

GView Online

Page 9: Citethis:Phys. Chem. Chem. Phys.,2011,13 ,61456155 …homepage.univie.ac.at/hans.lischka/pub_domain/Cytosine_dyn_PCCP_… · This ournal is c the Owner Societies 2011 Phys. Chem.

This journal is c the Owner Societies 2011 Phys. Chem. Chem. Phys., 2011, 13, 6145–6155 6153

state. Initially, cytosine relaxes along the pp* state quickly

reaching a region of strong mixing of the pp*, np* and closed

shell states. In this first approach to the ground state, 16% of

the trajectories are deactivated within a 13 fs time constant

(Fig. 9-1). Although the semi-planar minimum on the crossing

seam mixes the nOp* and the pp* states (Table 2), the actual

transitions took place in higher energy regions of the crossing

seam mainly involving the pp* and the closed shell states. Due

to limitations of surface hopping method, the value for this

very short time constant should be taken cautiously. Note,

however, that this semi-instantaneous deactivation of

cytosine through the semi-planar conical intersection was also

observed in the multiple spawning dynamics performed at

CASSCF(2,2) level.24 On the other hand, it was not observed

in surface hopping dynamics simulations at the semiempirical

OM2 level.30 In this case, the very short time constant reported

(0.04 ps) corresponds to a S2 - S1 deactivation process.

(A comparative summary of the results obtained in different

dynamics simulations of cytosine is given in Table 5.)

The remaining 84% of trajectories that did not deactivate in

this first encounter with the ground state relaxes to the nOp*state (Fig. 9-2). A second time constant of 0.69 ps appears

in the simulations corresponding to the time for internal

conversion of these trajectories. Three different reaction

pathways were observed at this time scale. The main one

was again the deactivation at the semi-planar region of the

crossing seam, which could occur every time the motion of the

molecule in the nOp* well brings it near this intersection. This

was the pathway followed by 52% of the trajectories (Fig. 9-3).

This finding is in strong contrast with the surface hopping

dynamics performed at OM2 level30 and the multiple spawning

dynamics performed at the CASSCF(2,2) level,24 where no

deactivation in this region of the crossing seam was observed

at the long time scale. The results of surface hopping dynamics

at CASSCF(12,9) are similar to the present results and show

predominance of the semi-planar conical intersection.85

The sloped shape of the conical intersection at this region

decreases its efficiency and opens the possibility of internal

conversion through other channels. A total of 15% of

trajectories escaped from the nOp* well by crossing the barrier

to biradical pp* states. Eight percent of them deactivated at

the C6-puckered conical intersection (Fig. 9-4). Other seven

percent deactivated at the oop-NH2 conical intersection

(Fig. 9-5). Again the present results are not in agreement with

previous dynamics simulations. While the oop-NH2 conical

intersection was the most accessed conical intersection in the

Fig. 8 Average potential energies during the dynamics simulations

computed at RI-CC2/SVP level. The inset shows the average C2–N3

bond length as a function of time. The gray area indicates the standard

deviation.

Fig. 9 Reaction pathways and time constants for dynamics of cytosine in the excited state.

Dow

nloa

ded

by T

exas

Tec

hnic

al U

nive

rsity

on

04 M

ay 2

011

Publ

ishe

d on

24

Febr

uary

201

1 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/C

0CP0

1327

GView Online

Page 10: Citethis:Phys. Chem. Chem. Phys.,2011,13 ,61456155 …homepage.univie.ac.at/hans.lischka/pub_domain/Cytosine_dyn_PCCP_… · This ournal is c the Owner Societies 2011 Phys. Chem.

6154 Phys. Chem. Chem. Phys., 2011, 13, 6145–6155 This journal is c the Owner Societies 2011

CASSCF(2,2) multiple spawning simulations, the C6-puckered

was the unique conical intersection observed in the OM2

surface hopping simulations (Table 5).

The remaining 17% of trajectories did not deactivate within

the 1.2 ps simulation time. They may still be part of the t2exponential decay or at least a fraction of them may decay

with a third and longer time constant of 3 ps. The duration

of our dynamics simulation is not sufficiently extended to

distinguish between the two possibilities. If the longer decay

time constant is assumed, the previous discussion is still valid,

but the numerical values for the time constants are slightly

modified to t1 = 9 fs and t2 = 0.53 ps (Table 4).

Different from our results, the analysis of reaction pathways

performed in ref. 42 at CASPT2 level disfavored the

semi-planar intersection and indicated a dominance of the

C6-puckered and oop-NH2 conical intersections. However,

comparison of the energy of the semi-planar conical inter-

section not only with present results based on full optimiza-

tion, but also with those of other CASPT2 and MRCI

investigations44,46 shows that the semi-planar structure is

probably computed too high in energy in ref. 42.

The different results obtained by several simulations are

expression of the discrepancies between the potential energy

surfaces computed at diverse levels. All of them predict

internal conversion occurring in sub-picosecond time scale

but with different populations of each deactivation pathway.

The CAS(2,2) employed in the multiple spawning dynamics is

likely a too small space to adequately describe the potential

energy surfaces of cytosine and it should be especially taken

with care for the description of the three state region of the

crossing seam, leading to a underestimation of the role of the

semi-planar conical intersection. Nevertheless, similar to our

results, it predicts a dominant trend of relaxation to the S1minimum from where cytosine escapes to different conical

intersections. In the case of the OM2 method, the lack of

activation of different conical intersections in any proportion

may be an indication of an overstabilization of the pathway

leading to the C6-puckered conical intersection. In comparison

to our results, a qualitatively different mechanism is predicted,

with direct deactivation along the pp* state without relaxing tothe S1 minimum. The CAS(14,10) active space used in our

simulations tends to predict the energy of the pp* state

relatively too high in comparison with the np* state. Because

of this, we believe that the proper stabilization of the pp* stateby inclusion of dynamical electron correlation should bring to

an increase of deactivation at the pp* channels.

4. Conclusions

We have investigated the photophysics of cytosine by non-

adiabatic dynamics simulations, optimization of stationary

points and conical intersections, and determination of reaction

paths. Optimizations have been performed at CASSCF and

MR-CISD levels and the dynamics simulations were performed

at CASSCF level with the surface hopping method including

four electronic states.

The results show the activation of multiple reaction

pathways in up to three different time scales, which correlates

well with the experimental results. Most of trajectories relax to

the np* S1 minimum. From this minimum, cytosine deactivates

mainly via a semi-planar conical intersection between the nOp*and the ground state in a region of the crossing seam near

a triple degeneracy. In fewer cases, it deactivates via two

different conical intersections involving crossings between

pp* states and the ground state. Dynamics of cytosine presents

a singular feature that is a semi-instantaneous internal

conversion of a minor fraction of the population within only

10 fs. The competition between reaction paths is controlled by

excited state barriers, and comparison to results of other

dynamics simulations shows that details of the potential

energy surfaces are important for the exact determination of

the role of each deactivation path.

Acknowledgements

This work has been supported by the Austrian Science Fund

within the framework of the Special Research Programs F41

(ViCoM) and of the German Research Foundation, Priority

Program SPP 1315, project No. GE 1676/1-1. This work was

part of the research project Z40550506 of the Institute of

Organic Chemistry and Biochemistry of the Academy of

Sciences of the Czech Republic. Support by the grant from

the Ministry of Education of the Czech Republic (Center for

Biomolecules and Complex Molecular Systems, LC512) and

Computer time at the Vienna Scientific Cluster (project nos.

70019 and 70151) is gratefully acknowledged.

References

1 L. Blancafort, J. Am. Chem. Soc., 2006, 128, 210.2 H. Chen and S. H. Li, J. Phys. Chem. A, 2005, 109, 8443.3 W. C. Chung, Z. G. Lan, Y. Ohtsuki, N. Shimakura, W. Domckeand Y. Fujimura, Phys. Chem. Chem. Phys., 2007, 9, 2075.

4 C. M. Marian, J. Chem. Phys., 2005, 122, 104314.5 S. Perun, A. L. Sobolewski and W. Domcke, J. Am. Chem. Soc.,2005, 127, 6257.

6 L. Serrano-Andres, M. Merchan and A. C. Borin, Proc. Natl.Acad. Sci. U. S. A., 2006, 103, 8691.

7 T. Gustavsson, A. Banyasz, E. Lazzarotto, D. Markovitsi,G. Scalmani, M. J. Frisch, V. Barone and R. Improta, J. Am.Chem. Soc., 2006, 128, 607.

8 S. Perun, A. L. Sobolewski and W. Domcke, J. Phys. Chem. A,2006, 110, 13238.

9 G. Zechmann and M. Barbatti, J. Phys. Chem. A, 2008, 112, 8273.10 M. Z. Zgierski, S. Patchkovskii, T. Fujiwara and E. C. Lim,

J. Phys. Chem. A, 2005, 109, 9384.11 N. Ismail, L. Blancafort, M. Olivucci, B. Kohler and M. A. Robb,

J. Am. Chem. Soc., 2002, 124, 6818.12 S. Matsika, J. Phys. Chem. A, 2004, 108, 7584.13 C. M. Marian, J. Phys. Chem. A, 2007, 111, 1545.14 M. Barbatti, A. J. A. Aquino, J. J. Szymczak, D. Nachtigallova,

P. Hobza and H. Lischka, Proc. Natl. Acad. Sci. U. S. A., 2010,107, 21453.

15 S. Perun, A. L. Sobolewski andW. Domcke,Mol. Phys., 2006, 104,1113.

16 L. Blancafort, B. Cohen, P. M. Hare, B. Kohler and M. A. Robb,J. Phys. Chem. A, 2005, 109, 4431.

17 S. B. Nielsen and T. I. Solling, ChemPhysChem, 2005, 6, 1276.18 M. Barbatti and H. Lischka, J. Phys. Chem. A, 2007, 111, 2852.19 J. A. Frey, R. Leist, C. Tanner, H. M. Frey and S. Leutwyler,

J. Chem. Phys., 2006, 125, 114308.20 E. Fabiano and W. Thiel, J. Phys. Chem. A, 2008, 112, 6859.21 M. Barbatti and H. Lischka, J. Am. Chem. Soc., 2008, 130, 6831.22 H. R. Hudock, B. G. Levine, A. L. Thompson, H. Satzger,

D. Townsend, N. Gador, S. Ullrich, A. Stolow andT. J. Martinez, J. Phys. Chem. A, 2007, 111, 8500.

Dow

nloa

ded

by T

exas

Tec

hnic

al U

nive

rsity

on

04 M

ay 2

011

Publ

ishe

d on

24

Febr

uary

201

1 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/C

0CP0

1327

GView Online

Page 11: Citethis:Phys. Chem. Chem. Phys.,2011,13 ,61456155 …homepage.univie.ac.at/hans.lischka/pub_domain/Cytosine_dyn_PCCP_… · This ournal is c the Owner Societies 2011 Phys. Chem.

This journal is c the Owner Societies 2011 Phys. Chem. Chem. Phys., 2011, 13, 6145–6155 6155

23 G. Groenhof, L. V. Schafer, M. Boggio-Pasqua, M. Goette,H. Grubmuller and M. A. Robb, J. Am. Chem. Soc., 2007, 129,6812.

24 H. R. Hudock and T. J. Martinez, ChemPhysChem, 2008, 9,2486.

25 J. J. Szymczak, M. Barbatti, J. T. Soo Hoo, J. A. Adkins,T. L. Windus, D. Nachtigallova and H. Lischka, J. Phys. Chem.A, 2009, 113, 12686.

26 H. Langer, N. L. Doltsinis and D. Marx, ChemPhysChem, 2005, 6,1734.

27 Y. Lei, S. Yuan, Y. Dou, Y. Wang and Z. Wen, J. Phys. Chem. A,2008, 112, 8497.

28 R. Mitric, U. Werner, M. Wohlgemuth, G. Seifert andV. Bonacic-Koutecky, J. Phys. Chem. A, 2009, 113, 12700.

29 Z. G. Lan, E. Fabiano and W. Thiel, ChemPhysChem, 2009, 10,1225.

30 Z. Lan, E. Fabiano and W. Thiel, J. Phys. Chem. B, 2009, 113,3548.

31 A. N. Alexandrova, J. C. Tully and G. Granucci, J. Phys. Chem. B,2010, 114, 12116.

32 H. Kang, K. T. Lee, B. Jung, Y. J. Ko and S. K. Kim, J. Am.Chem. Soc., 2002, 124, 12958.

33 C. Canuel, M. Mons, F. Piuzzi, B. Tardivel, I. Dimicoli andM. Elhanine, J. Chem. Phys., 2005, 122, 074316.

34 S. Ullrich, T. Schultz, M. Z. Zgierski and A. Stolow, Phys. Chem.Chem. Phys., 2004, 6, 2796.

35 K. Kosma, C. Schroter, E. Samoylova, I. V. Hertel and T. Schultz,J. Am. Chem. Soc., 2009, 131, 16939.

36 M. Daniels and W. Hauswirth, Science, 1971, 171, 675.37 J. W. Longworth, R. O. Rahn and R. G. Shulman, J. Chem. Phys.,

1966, 45, 2930.38 R. J. Malone, A. M. Miller and B. Kohler, Photochem. Photobiol.,

2003, 77, 158.39 L. Blancafort, M. J. Bearpark and M. A. Robb, in Radiation

Induced Molecular Phenomena in Nucleic Acid, ed. M. K. Shuklaand J. Leszczynski, Springer, Netherlands, 2008.

40 M. Z. Zgierski, T. Fujiwara and E. C. Lim, Chem. Phys. Lett.,2008, 463, 289.

41 K. Tomic, J. Tatchen and C. M. Marian, J. Phys. Chem. A, 2005,109, 8410.

42 L. Blancafort, Photochem. Photobiol., 2007, 83, 603.43 M. Merchan, R. Gonzalez-Luque, T. Climent, L. Serrano-Andres,

E. Rodriuguez, M. Reguero and D. Pelaez, J. Phys. Chem. B, 2006,110, 26471.

44 M.Merchan and L. Serrano-Andres, J. Am. Chem. Soc., 2003, 125,8108.

45 K. A. Kistler and S. Matsika, J. Chem. Phys., 2008, 128, 215102.46 K. A. Kistler and S. Matsika, J. Phys. Chem. A, 2007, 111, 2650.47 L. Blancafort andM. A. Robb, J. Phys. Chem. A, 2004, 108, 10609.48 A. L. Sobolewski and W. Domcke, Phys. Chem. Chem. Phys.,

2004, 6, 2763.49 J. C. Tully, Faraday Discuss., 1998, 110, 407.50 W. J. Hehre, R. Ditchfield and J. A. Pople, J. Chem. Phys., 1972,

56, 2257.51 R. Shepard, H. Lischka, P. G. Szalay, T. Kovar and M. Ernzerhof,

J. Chem. Phys., 1992, 96, 2085.52 R. Shepard, in Modern Electronic Structure Theory, ed.

D. R. Yarkony, World Scientific, Singapore, 1995, vol. 1, p.345.

53 H. Lischka, M. Dallos and R. Shepard, Mol. Phys., 2002, 100,1647.

54 H. Lischka, M. Dallos, P. G. Szalay, D. R. Yarkony andR. Shepard, J. Chem. Phys., 2004, 120, 7322.

55 M. Dallos, H. Lischka, R. Shepard, D. R. Yarkony andP. G. Szalay, J. Chem. Phys., 2004, 120, 7330.

56 W. C. Swope, H. C. Andersen, P. H. Berens and K. R. Wilson,J. Chem. Phys., 1982, 76, 637.

57 J. Butcher, J. Assoc. Comput. Mach., 1965, 12, 124.58 J. Pittner, H. Lischka and M. Barbatti, Chem. Phys., 2009, 356,

147.59 G. Granucci and M. Persico, J. Chem. Phys., 2007, 126, 134114.60 J. C. Tully, J. Chem. Phys., 1990, 93, 1061.61 S. Hammes-Schiffer and J. C. Tully, J. Chem. Phys., 1994, 101,

4657.62 D. Cremer and J. A. Pople, J. Am. Chem. Soc., 1975, 97, 1354.63 J. C. A. Boeyens, J. Chem. Crystallogr., 1978, 8, 317.64 V. Subramanian, K. Chitra, K. Venkatesh, S. Sanker and

T. Ramasami, Chem. Phys. Lett., 1997, 264, 92.65 M. Barbatti, A. J. A. Aquino and H. Lischka, Phys. Chem. Chem.

Phys., 2010, 12, 4959.66 A. Sharonov, T. Gustavsson, V. Carre, E. Renault and

D. Markovitsi, Chem. Phys. Lett., 2003, 380, 173.67 L. B. Clark, G. G. Peschel and I. Tinoco, J. Phys. Chem., 1965, 69,

3615.68 O. Christiansen, H. Koch and P. Jorgensen, Chem. Phys. Lett.,

1995, 243, 409.69 C. Hattig and F. Weigend, J. Chem. Phys., 2000, 113, 5154.70 C. Hattig and A. Kohn, J. Chem. Phys., 2002, 117, 6939.71 A. Bunge, J. Chem. Phys., 1970, 53, 20.72 S. R. Langhoff and E. R. Davidson, Int. J. Quantum Chem., 1974,

8, 61.73 P. J. Bruna, S. D. Peyerimhoff and R. J. Buenker, Chem. Phys.

Lett., 1980, 72, 278.74 H. Lischka, R. Shepard, F. B. Brown and I. Shavitt, Int. J.

Quantum Chem., 1981, S15, 91.75 H. Lischka, R. Shepard, R. M. Pitzer, I. Shavitt, M. Dallos,

T. Muller, P. G. Szalay, M. Seth, G. S. Kedziora, S. Yabushitaand Z. Y. Zhang, Phys. Chem. Chem. Phys., 2001, 3, 664.

76 H. Lischka, R. Shepard, I. Shavitt, R. M. Pitzer, M. Dallos,T. Muller, P. G. Szalay, F. B. Brown, R. Ahlrichs, H. J. Boehm,A. Chang, D. C. Comeau, R. Gdanitz, H. Dachsel, C. Ehrhardt,M. Ernzerhof, P. Hochtl, S. Irle, G. Kedziora, T. Kovar,V. Parasuk, M. J. M. Pepper, P. Scharf, H. Schiffer,M. Schindler, M. Schuler, M. Seth, E. A. Stahlberg, J.-G. Zhao,S. Yabushita, Z. Zhang, M. Barbatti, S. Matsika, M. Schuurmann,D. R. Yarkony, S. R. Brozell, E. V. Beck, J.-P. Blaudeau,M. Ruckenbauer, B. Sellner, F. Plasser and J. J. Szymczak,COLUMBUS, an ab initio electronic structure program, release5.9.2, 2008, www.univie.ac.at/columbus.

77 M. Barbatti, G. Granucci, M. Persico, M. Ruckenbauer,M. Vazdar, M. Eckert-Maksic and H. Lischka, J. Photochem.Photobiol., A, 2007, 190, 228.

78 M. Barbatti, G. Granucci, M. Ruckenbauer, J. Pittner, M. Persicoand H. Lischka, NEWTON-X: a package for Newtonian dynamicsclose to the crossing seam, 2007, www.newtonx.org.

79 R. Ahlrichs, M. Bar, M. Haser, H. Horn and C. Kolmel, Chem.Phys. Lett., 1989, 162, 165.

80 A. L. Spek, J. Appl. Crystallogr., 2003, 36, 7.81 C. Angeli, J. Comput. Chem., 2009, 30, 1319.82 A. L. Sobolewski and W. Domcke, Chem. Phys., 2000, 259, 181.83 S. Perun, A. L. Sobolewski and W. Domcke, Chem. Phys., 2005,

313, 107.84 K. Kosma, S. A. Trushin, W. Fuss and W. E. Schmid, J. Phys.

Chem. A, 2008, 112, 7514.85 J. Gonzalez-Vazquez and L. Gonzalez, ChemPhysChem, 2010, 11,

3617.

Dow

nloa

ded

by T

exas

Tec

hnic

al U

nive

rsity

on

04 M

ay 2

011

Publ

ishe

d on

24

Febr

uary

201

1 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/C

0CP0

1327

GView Online