You are on page 1of 10

Chemical Physics 376 (2010) 46–55

Contents lists available at ScienceDirect

Chemical Physics
journal homepage: www.elsevier.com/locate/chemphys

Electron flux during pericyclic reactions in the tunneling limit: Quantum


simulation for cyclooctatetraene
Hans-Christian Hege a, Jörn Manz b,⇑, Falko Marquardt a,c, Beate Paulus b, Axel Schild b
a
Abteilung Visualisierung und Datenanalyse, Zuse-Institut Berlin, Takustr. 9, 14195 Berlin, Germany
b
Institut für Chemie und Biochemie, Freie Universität Berlin, Takustr. 3, 14195 Berlin, Germany
c
Institut für Mathematik, Freie Universität Berlin, Arnimallee 6, 14195 Berlin, Germany

a r t i c l e i n f o a b s t r a c t

Article history: Pericyclic rearrangement of cyclooctatetraene proceeds from equivalent sets of two reactants to two
Received 30 June 2010 products. In the ideal limit of coherent tunneling, these reactants and products may tunnel to each other
In final form 30 July 2010 by ring inversions and by double bond shifting (DBS). We derive simple cosinusoidal or sinusoidal time
Available online 5 August 2010
evolutions of the bond-to-bond electron fluxes and yields during DBS, for the tunneling scenario. These
Dedicated to Professor Horst Köppel on the overall yields and fluxes may be decomposed into various contributions for electrons in so called pericy-
occasion of his 60th birthday. clic, other valence, and core orbitals. Pericyclic orbitals are defined as the subset of valence orbitals which
describe the changes of Lewis structures during the pericyclic reaction. The quantum dynamical results
Keywords: are compared with the traditional scheme of fluxes of electrons in pericyclic orbitals, as provided by
Electron flux arrows in Lewis structures.
Pericyclic reaction Ó 2010 Elsevier B.V. All rights reserved.
Tunneling
Cyclooctatetraene
Quantum reaction dynamics

1. Introduction pericyclic reaction in the tunneling domain will answer these


questions.
The purpose of this paper is to present first quantum dynamics For reference, electron-wavepacket reaction dynamics in the
simulations for electron fluxes during pericyclic reactions from pericyclic proton transfer of formamide has been analyzed by
reactants (R) to products (P) and back, in the domain of coherent Nagashima and Takatsuka, by means of their semiclassical
tunneling. The results shall be compared with the traditional rule Ehrenfest theory which contains single and double electronic exci-
for electronic and nuclear rearrangements of neighboring bonds tations, that means for energies much higher than the tunneling
which are graphically symbolized by curved arrows in Lewis struc- domain [19], see also [20,21].
tures of R, see the textbooks of organic chemistry [1–6], inorganic As a case study, we shall present applications to the pericyclic
chemistry [7,8], and biochemistry [9,10]. This rule allows simple rearrangement of cyclooctatetraene (COT) by tunneling. The Lewis
predictions and rationalizations of syntheses [11]. Previous structures of R and P with curved arrows are shown in Fig. 1; they
theoretical investigations have already focused on various related may be distinguished from each other by partial deuteration (not
aspects, such as the Woodward–Hoffmann rules for the conserva- shown). Both R and P have two conformations denoted R1, R2
tion of orbital symmetry [12], analysis of transition structures or and P1, P2, respectively, with equivalent tub-shaped non-aromatic
intermediates [13–16], and assignments of concerted versus structures of D2d symmetry, which may be interconverted by ring
sequential mechanisms [17,18]. With full respect of these achieve- inversion (RI) [14,22–32]. That means that COT has four equivalent
ments [1–18], several questions remain: How many electrons actu- isomers, R1, R2, P1 and P2. Irrespective of the two conformations
ally flow? In which direction do they flow? Which orbitals are R1 and R2, the traditional rule suggests to use the Lewis structure
involved? Just those – we call them pericyclic orbitals – which of R with double-headed or two single-headed curved arrows
are related to the bond pattern differences between the Lewis pointing from carbon–carbon double bonds C@C to one or two
structures of R and P? Or are other valence or core electrons in- neighboring single bonds CAC, respectively. For reference, each
volved as well? The present quantum dynamics simulation of a of the double- or single-headed arrows may be assigned to trans-
fers of two or one electrons, respectively. At the same time, nuclear
motions extend or compress the previous C@C and CAC bond
⇑ Corresponding author. Tel.: +49 30 838 53338; fax: +49 30 838 54792. lengths to those of CAC and C@C bonds, respectively, resulting in
E-mail address: jmanz@chemie.fu-berlin.de (J. Manz). double bond shifting (DBS), combined with RI. Three possible sets

0301-0104/$ - see front matter Ó 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.chemphys.2010.07.033
H.-C. Hege et al. / Chemical Physics 376 (2010) 46–55 47

illustrated schematically in Fig. 2, compare with Refs. [14,30,31].


For convenience, the value of the PES of the four minima will be
set equal to zero. The energy gap to the first excited singlet state
S1(11A2) is large (ca 386.2 kJ/mol [31]), so that the wavefunctions
describing R1, R2, P1, P2 may well be expressed by means of the
BOA. Each of the four equivalent structures, for example R1, can
be converted into all the others, e.g. into R2, P1 and P2. Specifically,
two equivalent ring inversions RIR and RIP between the reactants
R1 and R2 as well as between the products P1 and P2, may proceed
via the corresponding transition states TSRRI and TSPRI, respec-
tively. These have D4h symmetry. Likewise, DBS converts TSRRI
via the transition state TSDBS with D8h symmetry into TSPRI. The
corresponding levels of the first excited doublet of states repre-
senting tunneling by DBS have been assigned by means of transi-
tion state spectroscopy [14]. Accordingly, tunneling from R to P
is mode-selective, even for energies close to TSDBS, which is well
above the potential minima. For comparison, for energies that al-
low over-the-barrier or close-to-the barrier processes, the most
Fig. 1. Lewis structures of cyclooctatetraene (COT, C8H8) for reactant (R) and efficient paths from reactants to products may be shortcuts, which
product (P); R and P may be distinguished by partial deuteration (not shown). The lead, for example, from R1 to P2 via TSDBS but not necessarily via
curved double- or single-headed arrows symbolize transfers of two or one
TSRRI and TSPRI. Instead, they may lead to each other via two differ-
electrons, respectively. The top, bottom and middle versions on the left hand side
correspond to clockwise, counter-clockwise and pincer-motion-type fluxes of ent valley ridge inflections, VRIR and VRIP which are located be-
electrons during the pericyclic reaction. The simple schemes focus on the changes tween TSDBS and TSRRI as well as TSPRI [30,31]. Some of the
of the anti-aromatic electronic structure by double bond shifting (DBS). The possible paths between R1, R2, P1 and P2 are shown schematically
corresponding pairs of tub-shaped structures R1, R2 and P1, P2 with D2d
in Fig. 2. In contrast, the present coherent tunneling between R1,
symmetries representing R and P are shown in Fig. 2. The four shaded areas which
are arranged in the form of a Maltese cross are assigned to the four equivalent
R2, P1, P2, with energies well below the barriers, will be described
domains of double bonds for the reactants. DBS converts them into the corre- by means of the lowest quartet of delocalized vibrational eigen-
sponding four domains of single bonds for the products. Each domain has two states, without any explicit reference to these paths.
boundaries which consist of half planes that originate from the C2 symmetry axis The theory for electron fluxes during coherent tunneling in COT
and contain one of the carbon nuclei. These boundaries serve as observer planes for
is developed in Section 2. The results and discussions, and the con-
monitoring the electronic flux out of the four individual domains for the double
bonds. The curved arrows suggest that pericyclic rearrangement of COT is clusions are in Sections 3 and 4, respectively.
associated with the flux of 4  2 = 8 electrons.
2. Theory
of arrows in Lewis structures representing DBS of cyclooctatetra-
ene COT are shown in Fig. 1 – they symbolize different clockwise, 2.1. Concerted electronic and nuclear dynamics during pericyclic
counter-clockwise or pincer-motion-type electron fluxes which all rearrangement of COT in the Born–Oppenheimer approximation
lead to the same net pericyclic rearrangement of COT.
The electron flux during the pericyclic reaction will be evalu- The wavefunction representing pericyclic rearrangement of COT
ated using the approach of Ref. [33] with extensions; for alterna- in the electronic ground state is written in BOA as
tive recent approaches to electron fluxes during chemical
reactions, see [19–21,34]. Collective quantum tunneling of strongly Wðq; Q ; tÞ ¼ wel ðq; Q Þwnu ðQ ; tÞ: ð1Þ
correlated electrons in commensurate mesoscopic rings has been Here wel and wnu are the real time-independent electronic and the
investigated previously by Baldea, Köppel and Cederbaum [35], complex time-dependent nuclear wavefunctions, respectively,
see also Refs. [36,37]. The method [33] is based on the Born– depending on the electronic spatial and spin coordinates (q) and
Oppenheimer approximation (BOA). It has been developed and on the nuclear ones Q. In Dirac notation, wnu(Q, t) = hQjwnu(t)i. The
tested against rigorous non-BOA calculations, for electron fluxes corresponding notation for the spatial coordinates is r and R,
in vibrating Hþ 2 , see also Ref. [38]. The present extension of the respectively. For convenience, we shall assume that initially, at time
method is from Hþ 2 to the much larger molecule COT where quan- t = 0, the system is prepared in one of the configurations of the reac-
tum non-BOA simulations of electronic fluxes are prohibitive, and tants, say R1,
from coherent vibrations to tunneling. The importance of tunneling
during pericyclic reactions, even at room temperature, has been
Wðq; Q ; t ¼ 0Þ ¼ wel ðq; Q ÞwR1 ðQ Þ; ð2Þ
discovered already in Ref. [26]; tunneling during pericyclic rear- where wR1 denotes the nuclear wavefunction which is localized
rangement of COT has been anticipated in Ref. [29], albeit without close to the potential minimum for the reactant say R1, i.e., close
any explicit considerations of electron fluxes. For the present pur- to RR1. In the present application, the wavefunction W(q, Q, t) should
pose, coherent tunneling will be represented by a time dependent then describe coherent tunneling from R1 to all other conforma-
coherent nuclear wavefunction which is constructed in analogy to tions R2, P1, P2 of COT, as well as all the subsequent tunneling pro-
the text book example of coherent tunneling in a symmetrical dou- cesses, including revivals or partial revivals of R1. Analogous
ble well potential [39]. Tunneling between four equivalent isomers expressions may be derived if COT is prepared in one of the other
has been investigated previously in Ref. [40]. conformations R2, P1 or P2, represented by nuclear wavefunctions
For reference, it is helpful to summarize some important prop- wR2(Q), wP1(Q), wP2(Q) which are localized close to the potential
erties of COT, which have been discovered in the pioneering inves- minima at RR2, RP1 and RP2, respectively.
tigations [14,22–32]. It has a symmetric quadruple minimum The Hamiltonian representing COT is written as
potential energy surface (PES) V supporting the four equivalent iso-
H ¼ T nu þ Hel ; ð3Þ
mers R1, R2, P1 and P2, in the electronic ground state S0(11A1); the
set of nuclear coordinates (R) at the corresponding potential min- where Tnu accounts for the nuclear kinetic energy, and Hel is the
ima will be denoted by RR1, RR2, RP1 and RP2, respectively. This is electronic Hamiltonian
48 H.-C. Hege et al. / Chemical Physics 376 (2010) 46–55

Fig. 2. Topology of the symmetric quadruple minimum potential energy surface (PES) of cyclooctatetraene (COT) versus three coordinates (short-hand notation: RIR, RIP,
DBS) for ring inversions of the reactants R1, R2 (RIR) and products P1, P2 (RIP), and for double bond shifting (DBS) (schematic). The four minima of the PES support those four
equivalent tub-shaped non-aromatic structures R1, R2, P1, P2 of D2d symmetry; they may be distinguished by partial deuteration (not shown). Furthermore, the PES has two
transitions states TSRRI and TSPRI of D4h symmetry between R1 and R2 and between P1 and P2, respectively. The PES also has a transition state TSDBS with D8h symmetry
between TSRRI and TSPRI, and two valley ridge inflections VRIR and VRIP, which are located between TSDBS and TSRRI as well as TSPRI, respectively. For reference, various paths
represent over- or close-to-the-barriers transitions between the structures R1, R2, P1, P2, by RIR, RIP and by DBS, compare with Refs. [14,30,31]. Coherent tunneling between
R1, R2, R3, R4, with energies well below the barriers, is described by means of the lowest quartet of delocalized vibrational tunneling states, without any explicit reference to
these paths, see text and Fig. 3.

Hel ¼ T el þ V el;el þ V el;nu þ V nu;nu ; ð4Þ Integration over all nuclear coordinates, all electronic spin coordi-
nates and all spatial electronic coordinates but one yields the time
which includes the kinetic energy of the electrons, as well as the
dependent one-electron density qel,t(r, t), where r denotes its spatial
Coulomb interactions of the electrons (Vel,el), of the electrons and
coordinates. Likewise, for a given set of nuclear coordinates Q, inte-
nuclei (Vel,nu), and of the nuclei (Vnu,nu). All other contributions to
gration of the all-electron density qel(q; Q) over all electronic spin
H are neglected, for the present purpose. As a consequence, the elec-
coordinates and all spatial coordinates but one, yields the time-
tronic (ground state, 1S0) and nuclear spin states are conserved, in
independent one-electron density qel,Q(r; Q) depending on Q. Inte-
the present model. For fixed nuclei, the electronic time-indepen-
gration over the (conserved) nuclear spins yields the corresponding
dent Schrödinger equation
one-electron density qel,R(r; R).
Hel wel ðq; Q Þ ¼ V el ðQ Þwel ðq; Q Þ; ð5Þ
2.2. Analogy of coherent tunneling in symmetric double and quadruple
yields the electronic wavefunction wel(q; Q) and its energy in the
potentials
electronic ground state, denoted as potential energy surface (PES)
Vel(Q) for the nuclear motions. The nuclear wavefunction is propa-
Before we construct the nuclear wavefunction wnu(Q, t) describ-
gated as a solution of the nuclear time dependent Schrödinger
ing coherent tunneling from R1 to all other conformations of COT,
(nuTDSE) equation,
it is helpful to recall the derivation of the tunneling wavefunction
d e [39], written with
~ nu ðQ ; tÞ in a symmetric double well potential V
w
ih w ðQ ; tÞ ¼ Hnu wnu ðQ ; tÞ; ð6Þ
dt nu tilde, for reference. By analogy, the potential minima for the reac-
tant R e and for the product Pe (possibly distinguishable by isotopic
starting from the initial wavefunction wnu (Q, t = 0) = wR1(Q). The
substitution) are located at R ~ and R e , respectively. Likewise, the
nuclear Hamiltonian is e R e P
nuclear wavefunctions representing R e and P e with the lowest pos-
Hnu ¼ T nu þ V el ðQ Þ: ð7Þ sible energy e ~ ðQ Þ and w
E are written as w ~ ðQ Þ; they are localized
eR eP
The resulting all-electron-all-nuclear density is ~ and R
close to R ~ , respectively. These localized wavefunctions
eR eP
may be expressed in terms of the doublet of the delocalized gerade
qðq; Q ; tÞ ¼ jwel ðq; Q Þj2 jwnu ðQ ; tÞj2 ¼ qel ðq; Q Þqnu ðQ ; tÞ: ð8Þ (+) and ungerade () eigenfunctions [39]
H.-C. Hege et al. / Chemical Physics 376 (2010) 46–55 49

~ ðQ Þ ¼ p1ffiffiffi ðw
w ~ þ ðQ Þ þ w
~  ðQ ÞÞ; ð9Þ
w+(Q), w+(Q), w(Q) with eigenenergies E++, E+, E+, E,
eR 2 respectively,
~ ðQ Þ ¼ p1ffiffiffi ðw
w ~ þ ðQ Þ  w
~  ðQ ÞÞ; ð10Þ 1
eP 2 wR1 ðQ Þ ¼ ðw ðQ Þ þ wþ ðQ Þ þ wþ ðQ Þ þ w ðQ ÞÞ; ð21Þ
2 þþ
~ þ ðQ Þ and w
respectively. The w ~  ðQ Þ are obtained as eigenstates of 1
wR2 ðQ Þ ¼ ðwþþ ðQ Þ þ wþ ðQ Þ  wþ ðQ Þ  w ðQ ÞÞ; ð22Þ
the time-independent nuclear Schrödinger equation (nuTISE) 2
1
e nu w
~  ðQ Þ ¼ e ~  ðQ Þ; wP1 ðQ Þ ¼ ðwþþ ðQ Þ  wþ ðQ Þ  wþ ðQ Þ þ w ðQ ÞÞ; ð23Þ
H Ew ð11Þ 2
1
where e
E þ and e
E  are the levels of the lowest tunneling doublet. wP2 ðQ Þ ¼ ðwþþ ðQ Þ  wþ ðQ Þ þ wþ ðQ Þ  w ðQ ÞÞ: ð24Þ
2
Both wavefunctions (9), (10) have the same mean energy
  The first and second subscripts indicate gerade (+) or ungerade ()
~¼1 e
E Eþ þ e
E : ð12Þ symmetries with respect to RI and DBS, respectively. The localized
2 wavefunctions (21)–(24) and the delocalized eigenstates are illus-
The tunneling splitting is trated schematically in Fig. 3.
All wavefunctions (21)–(24) have the same mean energy,
De
E¼e
E  e
Eþ: ð13Þ
1
The solution of the nuTDSE (6), starting from the reactant wave- E¼ ðEþþ þ Eþ þ Eþ þ E Þ: ð25Þ
4
function (9), is obtained as Eq. (9) extended by time-dependent
The corresponding tunneling splittings and the relations to the tun-
coefficients
neling times are
~ ðQ ; tÞ ¼ p1ffiffiffi ðw
w ~ þ ðQ Þ expði e ~  ðQ Þ expði e
E þ t=hÞ þ w E  t=
hÞÞ: ð14Þ DEDBS ¼ Eþ  Eþþ  E  Eþ ð26Þ
eR 2
and
The resulting absolute square of the correlation function
     DEDBS sDBS ¼ h; ð27Þ
hw ~ R ðtÞi ¼ 1 exp i e
~ R ð0Þjw E þ t=h þ exp i e
E  t=h ; ð15Þ
2 for DBS as well as

yields the probability of observing the system as reactant DERI ¼ Eþ  Eþþ  E  Eþ ð28Þ
e ðtÞ ¼ jhw
P ~ ðtÞij ¼ cos2 ðD e
~ ð0Þjw 2
E pt=hÞ: ð16Þ and
e
R e e R R
DERI sRI ¼ h; ð29Þ
Likewise, the probability of observing the system as product is
for RI. We anticipate the relations
e ðtÞ ¼ jhw
P ~ ðtÞij2 ¼ sin2 ðD e
~ ð0Þjw E pt=hÞ: ð17Þ
e
P e e P R DEDBS  DERI ; ð30Þ
e ðtÞ and P
Obviously, P e ðtÞ oscillate periodically, with tunneling
eR eP or
time s
~ which is related to the tunneling splitting,
sDBS  sRI ; ð31Þ
De
Es~ ¼ h: ð18Þ
because of the different potential barriers for tunneling by DBS
The corresponding all-electron-all-nuclear density is compared to RI [28,30,31],

q~ ðq; Q ; tÞ ¼ jw~ el ðq; Q Þj2 jw~ eR ðQ ; tÞj2 ¼ q~ el ðq; Q Þq~ nu ðQ ; tÞ: ð19Þ V DBS > V RI ; ð32Þ
the most recent Ref. [31] suggests VDBS = 55.2 and VRI = 51.9 kJ/mol.
Integration of the all-electron density q ~ el ðq; Q Þ over all nuclear
The solution of the nuTDSE (6), starting from the reactant wave-
coordinates, all electronic spin coordinates and all spatial electronic
function (21), is
coordinates but one (dq ~ ) yields the one-electron density q ~ el;t ðr; tÞ.
Taking into account that the nuclear wavefunctions are centered 1
wR1 ðQ ; tÞ ¼ ðw ðQ ÞeiEþþ t=h þ wþ ðQ ÞeiEþ t=h
at Re with probability Pe ðtÞ and at Re with probability P e ðtÞ, we 2 þþ
R R P P
obtain the approximation þ wþ ðQ ÞeiEþ t=h þ w ðQ ÞeiE t=h Þ: ð33Þ
Z
q~ el;t ðr; tÞ ¼ q~ el ðq; Q Þq~ nu ðQ ; tÞdQ dq~ The resulting absolute square of the nuclear correlation function
Z h i 1
hwR1 ð0ÞjwR1 ðtÞi ¼ ðexpðiEþþ t=hÞ þ expðiEþ t=hÞ
 q~ el;R ðr; ReR Þ dðR  ReR Þ Pe eR ðtÞ þ dðR  ReP Þ PeeP ðtÞ dR 4
þ expðiEþ t=hÞ þ expðiE t=hÞÞ; ð34Þ
q e ðtÞ þ q
~ el;R ðr; Re Þ P e ðtÞ:
~ el;R ðr; ReÞ P
e
R R e P P
yields the probability of observing COT as reactant R1,
ð20Þ
PR1 ðtÞ ¼ jhwR1 ð0ÞjwR1 ðtÞij2
2.3. Nuclear tunneling dynamics of COT ¼ cos2 ðDEDBS pt=hÞ cos2 ðDERI pt=hÞ: ð35Þ
Likewise, the probabilities of observing COT as R2, P1 or P2 are
The derivation of the corresponding expressions for coherent
2
tunneling of COT is entirely analogous to Eqs. (1)–(20), except that PR2 ðtÞ ¼ jhwR2 ð0ÞjwR1 ðtÞij2 ¼ cos2 ðDEDBS pt=hÞ sin ðDERI pt=hÞ; ð36Þ
we have to consider four potential minima supporting two reac- 2 2 2
PP1 ðtÞ ¼ jhwP1 ð0ÞjwR1 ðtÞij ¼ sin ðDEDBS pt=hÞ cos ðDERI pt=hÞ; ð37Þ
tants R1, R2 and two products P1, P2 instead of just two minima
supporting fR and eP, respectively. The corresponding localized nu- and
clear wavefunctions wR1(Q), wR2(Q), wP1(Q), wP2(Q), are superposi- 2 2
tions of the lowest quartet of delocalized eigenstates w++(Q), PP2 ðtÞ ¼ jhwP2 ð0ÞjwR1 ðtÞij2 ¼ sin ðDEDBS pt=hÞ sin ðDERI pt=hÞ; ð38Þ
50 H.-C. Hege et al. / Chemical Physics 376 (2010) 46–55

Fig. 3. Topologies of the quartet of delocalized nuclear eigenstates w++(Q), w+(Q), w+(Q), w(Q) of COT, and the corresponding localized wave functions representing
reactants R1, R2 and products P1 and P2 (schematic). The bird’s eye view shows equicontours of the wavefunctions representing the eight carbon nuclei of COT. The
delocalized eigenstates have 32 lobes, which are centered at the four different positions for those eight carbon nuclei in configurations R1, R2, P1, P2. These 32 lobes are
arranged as 16 dumbbells, with front and back lobes which are shown as balls and partially hidden balls, respectively. The straight single and double lines indicate single and
double bonds of the Lewis structures of the reactants and products, respectively, to guide the eye from the delocalized wavefunctions to the localized ones. Dark and bright
lobes have positive and negative values, respectively, due to excitations of the nuclear motions for DBS (w+), ring inversion (w+), or both (w). The linear combinations
(21)–(24) imply destructive interferences which eliminate 32–8 lobes of the delocalized wavefunctions, to the benefit of constructive superpositions of eight lobes for the
localized wavefunctions WR1, WR2, WP1, WP2 representing the four different configurations R1, R2, P1, P2, respectively. The delocalized wavefunctions are arranged in
energetic order E++ < E+ < E+ < E. The corresponding set of localized wave functions have the same mean energies E, Eq. (25).

respectively. Gratifyingly, bonds of the reactant (DB, R), with equal volumes V ¼ V DB;R , to four
domains of single bonds of the products (SB, P), with corresponding
PR1 ðtÞ þ PR2 ðtÞ þ PP1 ðtÞ þ PP2 ðtÞ ¼ 1: ð39Þ equal volumes V ¼ V SB;P , see Fig. 1. The individual volumes have
corresponding boundaries (areas) A = ADB,R or A = ASB,P, respec-
The probability of finding the COT as reactant R, irrespective of the tively. These boundaries consist of two half planes that originate
configurations R1 or R2, is from the C2 symmetry axis such that each of them contains one
of two neighboring carbon nuclei which are connected by a
PR ðtÞ ¼ P R1 ðtÞ þ P R2 ðtÞ ¼ cos2 ðDEDBS pt=hÞ: ð40Þ carbon–carbon bond. The electronic flux out of one of these
domains with volume V through its boundary A is [33]
Likewise,
2
Z
PP ðtÞ ¼ PP1 ðtÞ þ PP2 ðtÞ ¼ sin ðDEDBS pt=hÞ; ð41Þ d
F el ðt; AÞ ¼ dV qel;t ðr; tÞ: ð43Þ
dt V
for the products. Expressions (40), (41) for COT are entirely analo-
gous to (16), (17) for the model system with a symmetric double
The time-dependent one-electron density qel,t(r, t) and the time-
well PES. Taking into account the relations (30), (31), comparison
independent one-electron density qel,R(r; R) are evaluated as de-
of expressions (35)–(38) and (40), (41) allow illuminating interpre-
scribed in the text after Eq. (8). By analogy with the derivation of
tations, e.g. slow tunneling from R to P by DBS is modulated by
expression (20) for the one-electron density q ~ el;t ðr; tÞ for tunneling
more rapid tunneling between R1 and R2 as well as between P1
in a symmetric double well potential, we take into account that
and P2, by means of RI. Averaging over very long times (indicated
the nuclear wavefunctions are centered at RR1, RR2, RP1 or RP2 with
by P) yields equal mean probabilities for observing COT in configu-
probabilities PR1(t), PR2(t), PP1(t), PP2(t), respectively, cf. Eqs. (35)–
rations R1, R2, P1, P2,
(38), thus
1
PR1 ¼ PR2 ¼ PP1 ¼ PP2 ¼ : ð42Þ qel;t ðr; tÞ  qel;R ðr; RR1 ÞPR1 ðtÞ þ qel;R ðr; RR2 ÞPR2 ðtÞ
4
þ qel;R ðr; RP1 ÞPP1 ðtÞ þ qel;R ðr; RP2 ÞPP2 ðtÞ: ð44Þ

2.4. Electron flux during pericyclic tunneling of COT Inserting the approximation (44) into (43) allows to express the
electronic flux (43) as the sum of two contributions which depend
For reference, the curved arrows in the Lewis structure of the on the nuclear configurations of the reactants (R) and products (P)
reactant R suggest that four double bonds of COT are converted
F el ðt; AÞ  F el;R ðt; AÞ þ F el;P ðt; AÞ: ð45Þ
to four single bonds of the products P, such that altogether 4  2
electrons are transferred from four equivalent domains of double Specifically, for the double bond domain V DB;R of the reactants
H.-C. Hege et al. / Chemical Physics 376 (2010) 46–55 51

Z Z !
d the absolute value of the negative yield (53) – is just the difference
F el;DB;R ðt; AÞ ¼ PR1 ðtÞ dV qel;R ðr; RR1 Þ þ P R2 ðtÞ dV qel;R ðr; RR2 Þ
dt V DB;R V DB;R
between the number of electrons in the four equivalent domains of
the double bonds of the reactants, with equal volumes V DB;R , minus
d 
 PR1 ðtÞN el;DB;R ðR1Þ þ PR2 ðtÞN el;DB;R ðR2Þ ; the number of electrons in the four equivalent domains of the nas-
dt
cent single bonds of the products, with equal volumes V SB;P . The
ð46Þ
time evolution of electron transfer from R to P during DBS by tun-
where Nel,DB,R(R1), Nel, DB,R(R2) denote the numbers of electrons in neling – that means four times the absolute value of the negative
the volume V DB;R of the domain of the reactants if the nuclei are yield (52) – is given by the simple cos-type expression, with tunnel-
localized at the configurations R1 or R2 of the reactants, respec- ing period sDBS, cf. Eq. (27). Finally, the flux of electrons out of the
tively. For symmetry reasons, these numbers are the same, i.e., RI four equivalent domains of the double bonds of the reactants, or
between R1 and R2 does not change the numbers of electrons in out of the corresponding domains of the nascent single bonds of
the domains of the double bonds of the reactants. Accordingly, the products, is simply four times the time derivative of the yield
using the unifying notation [39], expression (51), with the same tunneling period sDBS. RI does
Nel;DB;R ¼ Nel;DB;R ðR1Þ ¼ Nel;DB;R ðR2Þ; ð47Þ not affect these overall results for the electron flux during the peri-
cyclic reaction of COT in the tunneling domain. Nevertheless, both
together with Eq. (40), we obtain RI and DBS are important if one monitors the population of a spe-
cific reactant or product, say R1, cf. Eqs. (35)–(38).
d
F el;DB;R ðt; AÞ ¼ Nel;DB;R P R ðtÞ
dt
2.5. Transfer of electrons in pericyclic orbitals, other valence orbitals,
¼ Nel;DB;R ðDEDBS p=hÞ sinð2DEDBS pt=hÞ and core orbitals during the pericyclic rearrangement of COT in the
1 tunneling limit
¼  N el;DB;R ðDEDBS =hÞ sinðDEDBS t=hÞ: ð48Þ
2
Likewise, RI between the products P1 and P2 does not change the The one-electron densities qel,R(r; RR1), qel,R(r; RR2), qel,R(r; RP1),
numbers of electrons in any of the four domains of the carbon–car- qel,R(r; RP2) for the four different configurations R1, R2, P1 and P2
bon single bonds of the nascent products, with volume V SB;P . Hence, of COT are calculated with the MOLPRO program [41]. To obtain
using accurate results, a multi-reference configuration interaction
Z (MRCI) calculation with Davidson correction was performed on
Nel;SB;P ¼ N el;SB;P ðP1Þ ¼ Nel;SB;P ðP2Þ  dV qel;R ðr; RP1 Þ top of a CAS (8, 8) calculation including the electrons of p-charac-
V SB;R ter. A cc-pVTZ basis set was used [42]. The structure obtained after
Z
geometry optimization in D2d-symmetry compares very well with
¼ dV qel;R ðr; RP2 Þ; ð49Þ the experimental data of Ref. [32], with calculated structure param-
V SB;R
eters d(CAC) = 147.8 pm, d(C@C) = 134.0 pm, d(CAH) = 106.9 pm,
together with Eq. (41) yields \(CACAC) = 2.205 rad, \(HAC@C) = 2.058 rad and \(HACAC) =
d 2.018 rad. The corresponding angles between the intersecting
F el;SB;P ðt; AÞ ¼ Nel;SB;P PP ðtÞ boundaries for double and single bonds are \(CAC) = 0.753 and
dt
\(C@C) = 0.818 rad, respectively. The ratio of the volumes are thus
1
¼ Nel;SB;P ðDEDBS =hÞ sinðDEDBS t=hÞ; ð50Þ
2 V SB \ðC  CÞ
¼ ¼ 0:920; ð54Þ
for the contribution of the electronic flux during the pericyclic reac- V DB \ðC ¼ CÞ
tion R ? P, out of any of the four single bond domains of the nascent
products, with volume V ¼ V SB;P . Inserting (48) and (50) into (45) different from the ratio of the (tilted) bond lengths, d(CAC)/
yields the final result, d(C@C) = 1.103.
For further investigation, the densities may be decomposed into
1 three contributions: Electrons occupying ‘‘pericyclic” (peri) orbitals
F el ðt; AÞ  ðN el;SB;P  Nel;DB;R ÞðDEDBS =hÞ sinðDEDBS t=hÞ: ð51Þ
2 which correspond to the active orbitals of the CAS calculation,
The difference Nel,SB,P  Nel,DB,R accounts for the transformation of other valence (oval) orbitals, and core (core) electrons which are
the double bond of R into the single bond of P, during the pericyclic the 1s orbitals of the carbon atoms, for example
reaction. The resulting (negative) yield – that means loss – of the
electrons which leave the domain of the initial double bond is [39] qel;R ðr; RR1 Þ ¼ qel;R;peri ðr; RR1 Þ þ qel;R;oval ðr; RR1 Þ
Z t þ qel;R;core ðr; RR1 Þ: ð55Þ
0
Y el ðt; AÞ ¼ dt F el ðt 0 ; AÞ
0
1 Essentially, the pericyclic orbitals are the ones which correspond to
 ðNel;SB;P  Nel;DB;R Þð1  cosðDEDBS t=hÞÞ: ð52Þ different Lewis structures R and P of COT, cf. Fig. 1. The two contri-
2
butions for pericyclic and other valence electrons add up to the
The factor (1  cos(DEDBSt/⁄)) describes the effects of periodic electron density of the electrons in all the valence orbitals,
coherent tunneling R ? P ? R ? P ? etc., with period sDBS. In par-
ticular, for t = sDBS/2, when all the reactants R (=R1 or R2) have tun- qel;R;val ðr; RR1 Þ ¼ qel;R;peri ðr; RR1 Þ þ qel;R; oval ðr; RR1 Þ: ð56Þ
neled to products P (=P1 or P2) by DBS, the (negative) yield of
electrons which flow out of any of the four equivalent domains is Equivalent expressions apply to qel,R(r; RR2), qel,R(r; RP1) and
equal to the extreme value qel,R(r; RP2). All electron densities were calculated on a grid of
Y el ðsDBS =2; AÞ ¼ Nel;SB;P  Nel;DB;R : ð53Þ 440  440  440 points with grid spacing of 2.6 pm.
The decomposition (55) carries over to corresponding defini-
Eqs. (51)–(53) allow illuminating interpretations. Accordingly, the tions of the numbers of periyclic, other valence, core, or valence
maximum number of electrons which are transferred by tunneling electrons, e.g.
via DBS from R (irrespective of the conformations R1 or R2) to P
(irrespective of the conformations P1 or P2) – that means four times Nel;DB;R;k ¼ Nel;DB;R;k ðR1Þ ¼ Nel;DB;R;k ðR2Þ; ð57Þ
52 H.-C. Hege et al. / Chemical Physics 376 (2010) 46–55

for k = peri, oval, core, val cf. Eq. (47); to the resulting fluxes (c.f. Eq. 1
(51)),

PR1
1
F el;k ðt; AÞ  ðNel;SB;P;k  Nel;DB;R;k ÞðDEDBS =hÞ sinðDEDBS t=hÞ; ð58Þ 0
2
and to the (negative) yields (that means losses) of electrons which 1
flow out of the domain of the double bonds of the reactants, or out

PR2
of the domains of the single bonds of the nascent products, (cf. Eq.
(52)), 0

1 1
Y el;k ðt; AÞ  ðNel;SB;P;k  Nel;DB;R;k Þð1  cosðDEDBS t=hÞÞ: ð59Þ

P1
2

P
In particular, it will be interesting to use the results for the extreme
0
values of the electron yields (c.f. Eq. (53))
1
Y el;k ðsDBS =2; AÞ  Nel;SB;P;k  Nel;DB;R;k ; ð60Þ

PP2
for a quantification of the rules based on curved arrows in the Lewis
structure of the reactant: How many electrons flow out of each of 0
0 0.5 1 1.5
the four equivalent domains for double bonds of the reactant? t/τ
DBS

3. Results and discussions Fig. 4. Probabilities (a) PR1(t), (b) PR2(t), (c) PP1(t), (d) PP2(t) of observing the
reactants R1, R2, or the products P1 or P2, if COT has been prepared initially as R1,
In order to answer the questions which have been raised in Sec- cf. Eqs. (35)–(38), respectively. The envelope of PR1(t) and PR2(t) is the probability
PR(t) of observing the reactants R, Eq. (40). Likewise, the envelope of PP1(t) and PP2(t)
tion 1 about the electron fluxes during the pericyclic rearrange- is the probability PP(t) of observing the products P, Eq. (41). The slowly varying PR(t)
ment of COT in the tunneling limit, we first note that the validity and PP(t) indicate coherent tunneling R ? P ? R ? P ? etc. by DBS. The rapid
of the BOA implies that the electronic wavefunction wel(q,Q) of modulations of PR1(t) and PR2(t), or PP1(t) and PP2(t) indicate much faster coherent
the non-degenerate electronic ground state, Eq. (1), is real-valued. tunneling R1 ? R2 ? R1 ? R2 ? etc. or P1 ? P2 ? P1 ? P2 ? by RIR or RIP of the
reactants and products, respectively. The ratio of the tunneling times sDBS/sRI = 5.3
As a consequence, the mean value of electronic angular momen-
is adapted from the estimate (61).
tum for rotations around the C2 axis is zero. This is in accord exclu-
sively with the pincer-motion-type flux of electrons, which is
illustrated in the middle of the left hand side (lhs) of Fig. 1. In con- of Miller [47], see also Ref. [26]. For this purpose, we design exem-
trast, the familiar clockwise or counter-clockwise fluxes of elec- plary RI and DBS tunneling paths from the potential minimum rep-
trons, which are illustrated schematically in the top and bottom resenting R1 via VRI to R2, and from R1 via VDBS to P2, respectively.
versions of the lhs of Fig. 1, have opposite, non-zero mean values The lengths of the two tunneling paths are about equal, i.e., ca. 1.7
of angular momenta. Hence, they are ruled out for the pericyclic a0, cf. Fig. 2. For the barrier heights, we adapt the values VDBS = 55.2
rearrangement of COT in the electronic ground state. They remain and VRI = 51.9 kJ/mol [32], cf. Eq. (32). For comparison, the present
options, however, for electron fluxes in excited states, cf. Refs. MRCI calculation yields VDBS = 86.5 kJ/mol and VRI = 55.6 kJ/mol,
[19,43–46]. respectively. A detailed comparison of the results for various quan-
The representation of the pincer-motion-type flux of electrons tum chemical methods will be presented elsewhere. For a simple,
by eight single headed curved arrows in the Lewis structure of crude estimate, the profiles of the PES along these tunneling paths
COT (cf. Fig. 1) implies that out of each of the four domains of are fitted as symmetric polynomials Vtun,RI and Vtun,DBS of fourth or-
the double bonds, part of the pericyclic electrons flow clockwise der to these parameters. The reduced masses l for motions along
through one of the half plane boundaries, while the same amount these tunneling paths are the same, l = 2(mC + mH), because
of electrons flow counter-clockwise through the opposite half four fragments CH move versus the other four fragments CH during
plane boundary; see the definitions in the beginning of Section 2.4. RI as well as during DBS. The ratio of the tunneling times is then
Next we address the question of how many electrons actually [47]
flow. For this purpose, we evaluate the electron yields and fluxes,
cf. Eqs. (51)–(53) for all electrons as well as (58)–(60) for the elec- sDBS
¼ 5:3: ð61Þ
trons in pericyclic, other valence, all valence and in core orbitals. sRI
According to the derivation of Section 2, the results depend, in part,
on the probabilities PR(t) and PP(t) of observing COT as reactant R This ratio is used in Fig. 4, see the rapid modulations of the proba-
and product P, in the tunneling limit, cf. Eqs. (40), (41), (43)– bilities PR1(t) etc. due to RI, compared to the rather slow variation of
(53). These probabilities are illustrated in Fig. 4, together with the envelope representing PR(t), due to DBS. The subsequent results
the probabilities PR1(t), PR2(t) and PP1(t), PP2(t) of observing COT will be expressed in terms of sDBS, see Figs. 4, 6, irrespective of the
as reactants R1, R2, or products P1, P2, if the pericyclic rearrange- numerical value of sDBS. For example, sDBS = 1.9  106 s for the sim-
ment starts out from reactants R1. As anticipated, on one hand, ple model of Ref. [47]. More sophisticated theories of multidimen-
PR(t) and PP(t) display periodic, slow tunneling R ? P ? R ? sional tunneling may yield different values, but this is irrelevant
P ? etc. with tunneling time sDBS due to DBS, Eq. (27), without for the subsequent conclusions [48,49].
any effect of RI. On the other hand, the individual components The results for the electron fluxes and yields also depend on the
R1, R2 of R and P1, P2 of P show the same overall time evolutions difference of the number of electrons in one of the sectors for the
due to DBS as envelope, but this is modulated by the much faster double bonds of the reactants, minus the number of electrons for
tunneling by ring inversions of the reactant (RIR), R1 ? R2 ? the sector of the single bonds of the nascent products, see Eqs.
R1 ? R2 ? etc. as well as of the products (RIP) P1 ? P2 ? (51)–(53) for all electrons as well as Eqs. (58)–(60) for the parti-
P1 ? P2 ? etc., respectively, with tunneling time sRI, Eq. (29). tioning of the electrons into pericyclic, other valence, and core
The ratio of the tunneling times sRI/sDBS can be estimated, in electrons, see the derivation in Sections 2.4 and 2.5, respectively.
zero order approximation, by means of the semiclassical theory These numbers are evaluated, in principle, as integrals of the
H.-C. Hege et al. / Chemical Physics 376 (2010) 46–55 53

corresponding one-electron densities in the sectors for double (R) Nel;DB;R;peri  Nel;SB;P;peri ¼ 2Nel;DB;R;peri  8=4 ¼ þ0:37; ð66Þ
and single (P) bonds, respectively. The one-electron densities are Nel;DB;R;oval  Nel;SB;P;oval ¼ 2Nel;DB;R;oval  32=4 ¼ 0:86; ð67Þ
illustrated by transparent perspective three-dimensional (3d) equi-
Nel;DB;R;val  Nel;SB;P;val ¼ 2Nel;DB;R;val  40=4 ¼ 0:49; ð68Þ
density contour plots of the densities qel,k(r;RR1) and qel,k(r;RR2),
k = peri and val, for the reactants R1 and R2, in the upper and lower
Nel;DB;R;core  Nel;SB;P;core ¼ 2Nel;DB;R;core  16=4 ¼ 0:02: ð69Þ
left panels of Fig. 5, respectively. The upper and lower middle TThe values which are listed in Eqs. (65)–(69) are obtained by cor-
panels of Fig. 5 show corresponding bird’s eye views of the same responding integrations of the one electron densities qel,R(r;RR1) or
one-electron equidensities of R1 and R2, which can tunnel one to qel,R,k(r;RR1), k = peri,oval,val,core, in one of the sectors for double
another by RIR. Obviously, these bird’s eye views look the same, bonds of the reactant R, cf. Fig. 5. In fact, Eqs. (65)–(69) show that
due to the symmetry of R1 and R2. This confirms that the corre- in order to calculate the differences of the numbers of electrons in
sponding integrated numbers of electrons in one of the sectors the sectors for double bonds of the reactants minus the numbers
for double bonds are the same for R1 and R2, Eqs. (47), (57). Like- of electrons in the sectors for nascent single bonds of the products,
wise, Fig. 5 shows two bird’s eye views of the electron densities for readily for calculations of the electron fluxes and yields, cf. Eqs.
the products P1 and P2, with indications of the corresponding (51)–(53) and (58)–(60), all one needs is the corresponding num-
tunnelings by RIP, and by DBS. Again, they look the same due to bers of electrons in the sectors for double bonds, together with
the symmetry of P1 and P2. This confirms that the corresponding the total number of electrons or the overall number of electrons
integrated numbers of electrons in the sectors for the nascent sin- in the corresponding set of orbitals. The results (66)–(69) are con-
gle bonds are the same for P1 and P2, cf. Eq. (49). sistent with (65). The resulting time evolutions of the partial elec-
Obviously, the bird’s eye views of P1 and P2 look the same as tron yields Yel,k(t, A) as well as the related fluxes Fel,k(t, A), cf. Eqs.
those of R1 and R2, except for a rotation by p/4. This shows that (58)–(60), are illustrated in Fig. 6, for k = peri, oval and core.
the numbers Nel,SB,P or Nel,SB,P,k of the electrons in the sectors with Let us first discuss the results for the electron yields Yel,k(sDBS/
single bonds of the nascent products (subscript P) are the same as 2, A), starting with the electrons in pericyclic orbitals (k = peri),
the numbers of electrons in the corresponding sectors of single Yel,peri(s DBS/2, A)  0.37. Apparently, the absolute value of
bonds of the reactant (subscript R), Yel,peri(sDBS/2, A) is much smaller than the number 2 which one
may assign to the curved arrows in the Lewis structure, for refer-
Nel;SB;R ¼ N el;SB;P ; ð62Þ ence, see Fig. 1. The reason for this significant reduction from 2
Nel;SB;R;k ¼ Nel;SB;P;k for k ¼ peri; oval; val; core: ð63Þ to only j  0.37j is that the Lewis structure symbolizes, for simplic-
ity and for reference, the case when the electrons undergoing DBS
Eqs. (62), (63) yield simple relations of the numbers of electrons in are localized initially exclusively in the double bonds. In contrast,
sectors for single and double bonds. For example, COT has the total the present quantum simulations show that the electrons are delo-
number of electrons Nel = 56, hence calized not only in the domains of double bonds, but also in single
bonds, see Fig. 5. Considering the four double bonds of the reactant,
Nel ¼ 56 ¼ 4ðN el;DB;R þ N el;SB;P Þ; ð64Þ it suffices, therefore, that just 4  0.37 = 1.48, not 4  2 = 8 pericy-
clic electrons participate in double bond shifting. Another idea is to
thus compare the value 1.48 with the number of electrons in the highest
occupied molecular orbital (HOMO) which is related to DBS. How-
Nel;DB;R  Nel;SB;P ¼ 2Nel;DB;R  56=4 ¼ 0:47: ð65Þ ever, on one hand in post Hartree–Fock methods the concept of the
HOMO is not applicable, and on the other hand on Hartree–Fock le-
Likewise, the numbers of electrons in pericyclic, other valence, all vel this would still be an overestimation.
valence and in core electrons are Nperi = 8, Noval = 32, Nval = 40 and The simple model of curved arrows in Lewis structures does not
Ncore = 16. As a consequence, consider any electrons in other valence (k = oval) orbitals. In

Fig. 5. Equidensity contours for electrons in pericyclic orbitals (orange/ red, qel,R,peri = 0.04 and 0.4 a3 3
0 ), in all valence orbitals (blue/ green, qel,R,val = 0.27 and 0.53 a0 ) and for
all electrons (grey, qel,R = 0.05 a3
0 ), together with stick-and-ball plots of COT in the configurations of the reactants R1 and R2 (left and middle panels) and products P1 and P2
(right panels). The left panels show three-dimensional perspective views, whereas the middle and right panels show birds’ eye views. Also indicated are the two ring
inversions RIR and RIP of the reactants R and products P, as well as the double bond shifting DBS between R and P. The grey-shaded sectors correspond to one of the four
equivalent domains of double bonds (for R1) and single bonds (for P1), respectively (compare Fig. 1). The graphical representations have been generated using the AMIRA
package [50].
54 H.-C. Hege et al. / Chemical Physics 376 (2010) 46–55

coherent tunneling. The results are surprising, that means the net
Y flux of electrons goes from the domains of single bonds to double
0.6 el,core
bonds of the reactants, or to the nascent single bonds of the prod-
Y
ucts, not vice versa, as implied by the traditional arrows in Lewis
yield

el,oval

Yel,peri structures. The apparent paradox has been analyzed by decompo-


sition of the net flux into three components. It turns out that a
−0.6
dominant contribution stems from the r-bonds of the CH groups
which travel with the protons of the hydrogen atoms, or in general
it stems from other valence electrons which are not directly in-
Fel,core volved in double bond shifting of COT: all these components of
0.3
the net electron flux flow from the domains of single to double
F bonds. The second contribution is due to electrons in pericyclic
el,oval
flux

F
orbitals, that means those orbitals which are involved in the
el,peri
changes of the Lewis structures from reactants to products: These
−0.3
pericyclic components do indeed flow from the domains of double
to single bonds of the reactants, as symbolized by the arrows in Le-
0 0.5 1 1.5
t/τ wis structures. Quantitatively, the flux of pericyclic electrons out of
DBS
the four equivalent domains of double bonds is about as large – or
Fig. 6. (a) Partial yields Yel,k(t, A) of electrons in pericyclic (k = peri), other valence should one better say ‘‘as small” (=0.37  4 = 1.48) as the net
(k = oval) and core electrons (k = core), which leave one of the sectors for a double flux of electrons into these domains (=+0.47  4 = +1.88), but with
bond of COT through its boundary (A), cf. Eq. (59). (b) The corresponding partial opposite signs,  and +, respectively. The analogous effect, that
electron fluxes Fel,k(t, A), cf. Eq. (58).
means opposite fluxes of electrons in orbitals with p-character
compared to orbitals with r-character during pericyclic proton
contrast, quantum chemistry yields the result Yel,oval (sDBS/2, A)  transfer in formamide, at energies well above the tunneling do-
+0.86. Surprisingly, this contribution to the overall yield Yel(sDBS/ main, has been discovered first by Nagashima and Takatsuka [19].
2, A) is positive, i.e., the pericyclic electrons flow in the direction The quantitative results depend on the definition of the domain
opposite to the electrons in other valence orbitals. These trends boundaries. The sensitivity with respect to variation of the bound-
can be rationalized as follows. The electrons in the complementary aries will be discussed elsewhere. We anticipate that the results
valence electrons are dominated by the electrons occupying the will be robust for electrons in delocalized pericyclic and other
CH-r-bonds. Essentially, these travel from the domain of the single valence orbitals. In contrast, shifts of the boundaries away from
bonds of the reactants to the single bonds of the nascent products, the carbon nuclei may affect the results for electrons in core
cf. Figs. 1 and 4. orbitals, which are localized close to the nuclei.
As expected, the flux due to the electrons in the 1s electrons of It is a challenge to extend the present investigations to pericy-
the carbon nuclei (k = core) is negligible, Yel, core(sDBS/2,A)  0.02. clic rearrangements of COT at finite temperatures, comparable to
Summing up, the electron yield of all valence electrons is the experimental situations of Refs. [14,22–25,27,28,32]. As a
Y el;val ðsDBS =2; AÞ ¼ Y el;peri ðsDBS =2; AÞ þ Y el;oval ðsDBS =2; AÞ working hypothesis, the overall findings, which have been summa-
rized in the first paragraph of this conclusions section, will be ro-
¼ 0:37 þ 0:86 ¼ 0:49; ð70Þ bust, because they may be derived from exclusive quantum
i.e., the flux of pericyclic electrons is compensated by the antagonis- chemical evaluations of the one electron densities of the reactants
tic flux of all other valence electrons, mainly by the electrons occu- and products, irrespective of the quantum dynamical details. That
pying the 1s orbitals of the hydrogen atoms. Finally, the total means, for example, that we expect the same overall trends of
electron yield is antagonistic fluxes of electrons in pericyclic versus all other orbi-
tals, if the pericyclic rearrangement of COT proceeds at finite e.g.
Y el ðsDBS =2; AÞ ¼ Y el;val ðsDBS =2; AÞ þ Y el;core ðsDBS =2; AÞ room temperature.
¼ 0:49  0:02 ¼ þ0:47: ð71Þ It will also be a challenge to extend the present investigations to
pericyclic reactions of other systems. Will the trend of opposite
Accordingly, the total number (= 0.47) of electrons which are trans- directions of the fluxes of electrons in pericyclic versus all other
ferred during the pericyclic rearrangement of COT in the domain of orbitals be confirmed, again from the tunneling limit to reactions
coherent tunneling is about equal to the absolute value of the peri- say at room temperature? If so, a very important goal will be to
cyclic electrons (= j0.37j), but with opposite sign, i.e., the net flux gain a deeper understanding of the undoubted success of the rules
of electrons during DBS by tunneling is not out of, but into the do- of arrows in Lewis structures for predicting the course of pericyclic
mains of double bonds of the reactant. reactivity [1–11]. The present results already suggest a working
Fig. 6 shows the corresponding time evolutions of the electron hypothesis, i.e., pericyclic reactivity is governed by electrons in
yields and fluxes during pericyclic rearrangement of COT in the tun- pericyclic orbitals, irrespective of the even larger opposite fluxes
neling limit. As anticipated in Section 2.5, Eqs. (58) and (59), all the of electrons in other valence orbitals. The deeper understanding
individual Yel,k(t, A) and Fel, k(t, A) have the same time evolutions may well reveal a late triumph of the ingenious power of the sym-
(1  cos(t/sDBS) and sin(t/sDBS) depending on the tunneling time bolic arrows in Lewis structures, i.e., they focus on the essentials, at
by DBS, without any influence of RIR or RIP. The opposite directions least qualitatively. Work along these lines is in progress.
of the fluxes of the electrons in pericyclic orbitals compared to other
valence and core orbitals, as well as the quantitative dominance of Acknowledgements
the flux in other valence orbitals, are obvious from Fig. 6b.
We are very grateful to Professor H. Kono and Dr. K. Hoki
4. Conclusions (Tohoku University Sendai), Professor K. Takatsuka (Tokyo Univer-
sity) as well as Dr. R. Flesch (Freie Universität Berlin) for helpful
The present results for electron fluxes during pericyclic rear- discussions. We would also like to thank our colleagues from Or-
rangement of COT have been derived for the ideal situation of ganic Chemistry, Professors H. Ikeda (Osaka Prefecture University),
H.-C. Hege et al. / Chemical Physics 376 (2010) 46–55 55

D. Lenoir (Technische Universität München) and H.-U. Reissig [21] T. Yonehara, K. Takatsuka, Chem. Phys. 366 (2009) 115.
[22] F.A.L. Janet, J. Am. Chem. Soc. 84 (1962) 671.
(Freie Universität Berlin) for advice. Financial support from the
[23] F.A.L. Janet, A.J.R. Boum, Y.S. Lin, J. Am. Chem. Soc. 86 (1964) 3576.
Center of Scientific Simulations (Freie Universität Berlin), from [24] J.F.M. Oth, Pure Appl. Chem. 25 (1971) 573.
Deutsche Forschungsgemeinschaft (DFG, project Ma 515/25-1), [25] P.M. Thomas, A. Weber, J. Raman Spectrosc. 7 (1978) 353.
from the DFG research center MATHEON, and from Fonds der [26] M.J.S. Dewar, K.M. Merz Jr., J.J.P. Stewart, J. Am. Chem. Soc. 106 (1984)
4040.
Chemischen Industrie is also gratefully acknowledged. [27] M. Perec, Spectrochim. Acta 47A (1991) 799.
[28] D.A. Hrovat, W.T. Borden, J. Am. Chem. Soc. 114 (1992) 5879.
References [29] J.L. Andrés, O. Castaño, A. Morreale, P. Palmeiro, R. Gomperts, J. Chem. Phys.
108 (1998) 203.
[30] O. Castaño, R. Palmeiro, L.M. Frutos, J.L. Andrés, J. Comput. Chem. 23 (2002)
[1] J.A. Berson, P. de Mayo, Rearrangements in Ground and Excited States, vol. 1, 732.
Academic Press, New York, 1980. pp. 311.
[31] M. Garavelli, F. Bernardi, A. Cembran, O. Castaño, L.M. Frutos, M. Merchán, M.
[2] T.H. Lowry, K.S. Richardson, Mechanism and Theory in Organic Chemistry, Olivucci, J. Am. Chem. Soc. 124 (2002) 13770.
Harper and Row, New York, 1987. [32] D.S. Kummli, S. Lobsiger, H.-M. Frey, S. Leutwyler, J.F. Stanton, J. Phys. Chem. A
[3] J.J. Gajewski, Hydrocarbon Thermal Isomerizations, Elsevier, Amsterdam, 2004. 112 (2008) 9134.
[4] E.V. Anslyn, D.A. Dougherty, Modern Physical Organic Chemistry, University [33] I. Barth, H.-C. Hege, H. Ikeda, A. Kenfack, M. Koppitz, J. Manz, F. Marquardt, G.K.
Science Books, Sausalito, 2006.
Paramonov, Chem. Phys. Lett. 481 (2009) 118.
[5] M.B. Smith, J. March, Advanced Organic Chemistry Reactions, Mechanisms, and [34] P. von den Hoff, I. Znakovskaya, S. Zherebtsov, M.F. Kling, R. de Vivie-Riedle,
Structure, Wiley, Hoboken, 2007. Appl. Phys. B 98 (2010) 659.
[6] K.P.C. Vollhardt, N.E. Schore, Organic Chemistry – Structure and Function, [35] I. Baldea, H. Köppel, L.S. Cederbaum, Eur. Phys. J. B 20 (2001) 289.
Freeman, New York, 2007. [36] I. Baldea, H. Köppel, L.S. Cederbaum, Solid State Commun. 115 (2000) 593.
[7] R.G. Wilkins, Kinetics and Mechanism of Reactions of Transition Metal
[37] I. Baldea, H. Köppel, L.S. Cederbaum, Synth. Met. 119 (2001) 561.
Complexes, Verlag Chemie, Weinheim, 1991. [38] A. Kenfack, F. Marquardt, G.K. Paramonov, I. Barth, C. Lasser, B. Paulus, Phys.
[8] R.B. Jordan, Reaction Mechanisms of Inorganic and Organometallic Systems, Rev. A 81 (2010) 052502.
Oxford University Press, Oxford, 2007. [39] C. Cohen-Tannoudji, B. Diu, F. Laloë, Quantum Mechanics, Hermann, Paris,
[9] C.K. Mathews, K.E. van Holde, K.G. Ahern, Biochemistry, Prentice-Hall, Upper 1977.
Saddle River, 1999.
[40] O. Brackhagen, O. Kühn, J. Manz, V. May, R. Meyer, J. Chem. Phys. 199 (1994)
[10] D.L. Nelson, M.M. Cox, Lehninger Principles of Biochemistry, Freeman, New 9007.
York, 2009.
[41] H.-J. Werner, P.J. Knowles, et al., MOLPRO (version 2006.1), a package of ab
[11] G. Desimoni, G. Tacconi, A. Barco, G.P. Pollini, Natural Products Synthesis initio programs, see <http://www.molpro.net>.
Through Pericyclic Reactions, American Chemical Society, Washington, DC, [42] T.H.J. Dunning Jr., Chem. Phys. 90 (1989) 1007.
1983. [43] I. Barth, J. Manz, Angew. Chem. Int. Ed. 45 (2006) 2962.
[12] R.B. Woodward, R. Hoffmann, The Conservation of Orbital Symmetry, Verlag [44] I. Barth, J. Manz, Y. Shigeta, K. Yagi, J. Am. Chem. Soc. 128 (2006) 7043.
Chemie, Weinheim, 1970.
[45] M. Kanno, H. Kono, Y. Fujimura, Angew. Chem. Int. Ed. 45 (2006) 7995.
[13] K.N. Houk, Y. Li, J.D. Evanseck, Angew. Chem. Int. Ed. 31 (1992) 682. [46] M. Kanno, K. Hoki, H. Kono, Y. Fujimura, J. Chem. Phys. 127 (2007) 204314.
[14] P.G. Wenthold, D.A. Hrovat, W.T. Borden, W.C. Lineberger, Science 272 (1996) [47] W.H. Miller, J. Phys. Chem. 83 (1979) 960.
1456. [48] Y.-P. Liu, D.-h. Lu, A. Gonzalez-Lafont, D.G. Truhlar, B.C. Garrett, J. Am. Chem.
[15] O. Wiest, D.C. Montiel, K.N. Houk, J. Phys. Chem. A 101 (1997) 8378. Soc. 115 (1993) 7806.
[16] B.K. Carpenter, Angew. Chem. Int. Ed. 37 (1998) 3340.
[49] P.S. Zuev, R.S. Sheridan, T.V. Albu, D.G. Truhlar, D.A. Hrovat, W.T. Borden,
[17] M.J.S. Dewar, J. Am. Chem. Soc. 106 (1984) 209. Science 299 (2003) 867.
[18] L. Xu, C.E. Doubleday, K.N. Houk, J. Am. Chem. Soc. 132 (2010) 3029.
[50] D. Stalling, M. Westerhoff, H.-C. Hege, in: C.D. Hansen, C.R. Johnson (Eds.), The
[19] K. Nagashima, K. Takatsuka, J. Phys. Chem. A 113 (2009) 15240. Visualization Handbook, Elsevier, Amsterdam, 2005. chap. 38, pp. 749.
[20] M. Okuyama, K. Takatsuka, Chem. Phys. Lett. 476 (2009) 109.

You might also like