from orbitals to observables and backiopenshell.usc.edu/pubs/pdf/jcp_perspective.pdfquantum...
TRANSCRIPT
![Page 1: From orbitals to observables and backiopenshell.usc.edu/pubs/pdf/jcp_perspective.pdfQuantum mechanics explains chemical and physical phenomena by using abstract math-ematical constructs,](https://reader034.vdocuments.us/reader034/viewer/2022042406/5f1fa754210ddc366d060e82/html5/thumbnails/1.jpg)
From orbitals to observables and back
Anna I. Krylov
Department of Chemistry, University of Southern California, Los Angeles, California 90089-0482
Molecular orbital framework is of central importance in chemistry. Often used by
chemists and physicists to gain insight into molecular properties, Hartree–Fock or
Kohn–Sham orbitals are obtained from rather crude treatments and, strictly speak-
ing, are not observables. Yet, quantum mechanics offers a route for connecting
general many-electron wavefunctions with reduced quantities—density matrices and
orbitals—that give rise to observable properties. Such mapping makes possible,
in principle, reconstruction of these objects from sufficiently detailed experimental
data. This perspective discusses Dyson orbitals and various types of natural transi-
tion orbitals and illustrates their role in modeling and interpreting different types of
spectroscopic measurements.
I. INTRODUCTION
Quantum mechanics explains chemical and physical phenomena by using abstract math-
ematical constructs, such as point charges and masses, spins, Hamiltonians, and wavefunc-
tions. Today it is rightfully considered to be the foundation of chemistry, and most general
chemistry textbooks and courses begin with a brief foray into quantum mechanics. However,
back in the nineteenth century, this trait of quantum mechanics, its reliance on mathemat-
ical language and abstractions to describe physics, was not unanimously embraced by the
scientific community. Some of the best minds of the day denied quantum mechanics the
legitimacy of being a scientific discipline because it operates with objects that cannot be
observed.
Such was the opinion of one influential man from the 19th century, Ernst Mach1, a founder
of the Vienna Circle. Mach was not only an accomplished physicist (he even has a principle
named after him) but also a philosopher of science. The school of thought that he helped to
shape was called Logical Positivism, and one of its key principles was that science should be
based entirely on directly observable phenomena2. This sounds like something most of us
![Page 2: From orbitals to observables and backiopenshell.usc.edu/pubs/pdf/jcp_perspective.pdfQuantum mechanics explains chemical and physical phenomena by using abstract math-ematical constructs,](https://reader034.vdocuments.us/reader034/viewer/2022042406/5f1fa754210ddc366d060e82/html5/thumbnails/2.jpg)
2
subscribe to today, but there was a caveat in what Mach meant by observable phenomena.
Mach defended a type of phenomenalism that recognized only sensations as real. That is,
things you can see, hear, feel, or taste. Consequently, Mach considered atoms and molecules
to be artificial constructs of the mind. He famously declared after an 1897 lecture by Ludwig
Boltzmann at the Imperial Academy of Science in Vienna: “I don’t believe that atoms exist!”
These debates were eventually settled, and we now have a much broader view of what
observables are. We do not need to rely on our unaugmented faculties to interact with
the physical world. We accept that what we can see through microscopes, telescopes, or
spectrometers is real. But the main idea so ardently promoted by Mach is still valid—
that science should only deal with phenomena that can be observed. That is, from the
point of view of physical reality, a mathematical construct is only valid if one can design
an experiment that can, in principle, probe it. With this in mind, let us talk about a very
basic concept—orbitals. Are they real or just a mathematical construct?
When a common chemist hears “Quantum Mechanics”, she thinks “Orbitals”. Orbitals
are used to explain the structure of the periodic table of elements, the existence of chemical
bonds, trends in reactivity, and the colors of compounds. The role of molecular orbitals
in providing rational explanation of chemical transformations was recognized by the 1981
Nobel Prize given to Kenichi Fukui and Roald Hoffmann3.
Molecular orbital theory is now a truly native language of chemistry. Just as a small
child, without understanding its grammar and rules, learns to speak his native language
with great effectiveness, chemists learn how to speak orbitals before they learn quantum
mechanics. Indeed, some chemists never learn quantum mechanics, but use molecular orbital
theory with great effectiveness.
In this perspective, I wish to discuss molecular orbital theory with the experts, chemists
who not only speak the language, but also understand and appreciate its grammar. So let
us revisit the fundamentals and discuss the place of orbitals in quantitative and rigorous
physical chemistry.
In one-electron systems orbitals are wavefunctions. In a hydrogen-like atom, orbitals
correspond to its electronic states. In molecules, atomic orbitals give rise to delocalized
molecular orbitals, as shown in Fig. 1 for H+2 . The delocalization lowers the kinetic energy
of the electrons, gluing atoms into molecules. This is the quantum mechanical nature of
what we call a chemical bond4,5. Known as MO–LCAO theory (molecular orbital-linear
![Page 3: From orbitals to observables and backiopenshell.usc.edu/pubs/pdf/jcp_perspective.pdfQuantum mechanics explains chemical and physical phenomena by using abstract math-ematical constructs,](https://reader034.vdocuments.us/reader034/viewer/2022042406/5f1fa754210ddc366d060e82/html5/thumbnails/3.jpg)
3
1sA<latexit sha1_base64="yOMJ3odmWXYlYRxBbtl8aS4HVOU=">AAAB63icbVBNS8NAEJ3Ur1q/qh69LBbBU0mqoMeqF48V7Ae0oWy2m3bp7ibsboQS+he8eFDEq3/Im//GTZqDtj4YeLw3w8y8IOZMG9f9dkpr6xubW+Xtys7u3v5B9fCoo6NEEdomEY9UL8CaciZp2zDDaS9WFIuA024wvcv87hNVmkXy0cxi6gs8lixkBJtM8vTwZlituXU3B1olXkFqUKA1rH4NRhFJBJWGcKx133Nj46dYGUY4nVcGiaYxJlM8pn1LJRZU+2l+6xydWWWEwkjZkgbl6u+JFAutZyKwnQKbiV72MvE/r5+Y8NpPmYwTQyVZLAoTjkyEssfRiClKDJ9Zgoli9lZEJlhhYmw8FRuCt/zyKuk06t5FvfFwWWveFnGU4QRO4Rw8uIIm3EML2kBgAs/wCm+OcF6cd+dj0Vpyiplj+APn8weNX43q</latexit>
1sB<latexit sha1_base64="9l/3+z68asJRb9mqwP9BFJIvIRo=">AAAB63icbVBNS8NAEJ3Ur1q/qh69LBbBU0mqoMdSLx4r2A9oQ9lsN+3S3U3Y3Qgl9C948aCIV/+QN/+NmzQHbX0w8Hhvhpl5QcyZNq777ZQ2Nre2d8q7lb39g8Oj6vFJV0eJIrRDIh6pfoA15UzSjmGG036sKBYBp71gdpf5vSeqNIvko5nH1Bd4IlnICDaZ5OlRa1StuXU3B1onXkFqUKA9qn4NxxFJBJWGcKz1wHNj46dYGUY4XVSGiaYxJjM8oQNLJRZU+2l+6wJdWGWMwkjZkgbl6u+JFAut5yKwnQKbqV71MvE/b5CY8NZPmYwTQyVZLgoTjkyEssfRmClKDJ9bgoli9lZEplhhYmw8FRuCt/ryOuk26t5VvfFwXWu2ijjKcAbncAke3EAT7qENHSAwhWd4hTdHOC/Ou/OxbC05xcwp/IHz+QOO443r</latexit>
<latexit sha1_base64="uveq51XskeZ/BmBbyb/DEzkG8yU=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKexGQY9BLx4jmAckS5idzCZj5rHMzAphyT948aCIV//Hm3/jJNmDJhY0FFXddHdFCWfG+v63V1hb39jcKm6Xdnb39g/Kh0cto1JNaJMornQnwoZyJmnTMstpJ9EUi4jTdjS+nfntJ6oNU/LBThIaCjyULGYEWye1eoYNBe6XK37VnwOtkiAnFcjR6Je/egNFUkGlJRwb0w38xIYZ1pYRTqelXmpogskYD2nXUYkFNWE2v3aKzpwyQLHSrqRFc/X3RIaFMRMRuU6B7cgsezPxP6+b2vg6zJhMUkslWSyKU46sQrPX0YBpSiyfOIKJZu5WREZYY2JdQCUXQrD88ipp1arBRbV2f1mp3+RxFOEETuEcAriCOtxBA5pA4BGe4RXePOW9eO/ex6K14OUzx/AH3ucPna2PJw==</latexit>
<latexit sha1_base64="hGkufIW9kSvkyBAWr9V6gcNsI5A=">AAAB73icbVBNSwMxEJ2tX7V+VT16CRZBPJTdKuix6MVjBfsB7VqyabYNTbJrkhXK0j/hxYMiXv073vw3pts9aOuDgcd7M8zMC2LOtHHdb6ewsrq2vlHcLG1t7+zulfcPWjpKFKFNEvFIdQKsKWeSNg0znHZiRbEIOG0H45uZ336iSrNI3ptJTH2Bh5KFjGBjpU5Ps6HAD2f9csWtuhnQMvFyUoEcjX75qzeISCKoNIRjrbueGxs/xcowwum01Es0jTEZ4yHtWiqxoNpPs3un6MQqAxRGypY0KFN/T6RYaD0Rge0U2Iz0ojcT//O6iQmv/JTJODFUkvmiMOHIRGj2PBowRYnhE0swUczeisgIK0yMjahkQ/AWX14mrVrVO6/W7i4q9es8jiIcwTGcggeXUIdbaEATCHB4hld4cx6dF+fd+Zi3Fpx85hD+wPn8Abryj8M=</latexit>
FIG. 1: Orbital diagram explaining bonding in H+2 . Two atomic orbitals give rise to the in-phase
(σ) and out-of-phase (σ∗) molecular states, called molecular orbitals. The energy splitting between
σ and σ∗, which arises due to the different nodal structure (orbital shapes), explains chemical bond
formation.
combination of atomic orbitals), a qualitative generalization of this exact result is used to
explain trends in molecular properties at all levels, from freshman textbooks to state-of-the-
art research papers.
Being wavefunctions, orbitals of one-electron systems are connected to observables. Their
squares give the probability of finding an electron at a certain position in space, known as
electron density. Electron density can be probed directly by, for example, STM (scanning
tunneling microscopy), classic and ultrafast electron diffraction or X-ray scattering (see,
for example, Refs. 6,7). Different shapes of electron densities give rise to concrete prop-
erties, such as multipole moments. Differences in the shapes of orbitals corresponding to
different states encode optical properties—for example, one-photon transition between σ
and σ∗ orbitals in H+2 is induced only by light polarized along the molecular axis. Thus,
experiments that probe these properties probe the wavefunctions, and, consequently, orbital
shapes. Given sufficiently detailed information, orbital shapes can be reconstructed from
the experimental data. Hence, these orbitals are observables.
But what about many-electron systems? Their wavefunctions depend on the coordinates
of all electrons. They are explicitly non-separable and entangled. Orbitals are no longer
formally required to describe electronic states and their properties, rather, they are invoked
when discussing approximations to the true wavefunctions. For example, molecular orbitals
appear in the mean-field approximation, the Hartree–Fock model in which an N -electron
![Page 4: From orbitals to observables and backiopenshell.usc.edu/pubs/pdf/jcp_perspective.pdfQuantum mechanics explains chemical and physical phenomena by using abstract math-ematical constructs,](https://reader034.vdocuments.us/reader034/viewer/2022042406/5f1fa754210ddc366d060e82/html5/thumbnails/4.jpg)
4
many-body wavefunction is approximated by the antisymmetrized product of N orbitals
(called the Slater determinant)8:
Φ0 =1√N !P−1,2,3,...,n[φ1(1)φ2(2) . . . φN(n)] (1)
In this picture, orbitals describe states of pseudo-independent electrons, a very crude repre-
sentation of real correlated electrons. Even more troubling is that the key properties—energy
and electron density—do not depend on a specific choice of orbitals—any linear combination
of the occupied Hartree–Fock orbitals is equally good. So are orbitals just a mathematical
instrument devoid of concrete physical meaning?
The meaning of orbitals becomes even murkier when one considers strongly correlated
wavefunctions, which contain many Slater determinants, so that one can no longer rely
on the pseudo-independent electron model to assign a physical meaning to the individual
orbitals. So, one may wonder, do orbitals make any sense at all in many-electron systems?
In fact, orbitals do make sense and we can connect them to experimental observables.
And we can do this without invoking approximations—molecular orbital concepts are valid
within rigorous many-body formalisms. Molecular orbital theory is fully compatible with the
exact quantum mechanical description of many-electron systems, and orbitals are related to
physical observables, so that they are themselves observables. This contribution reviews the
formalism behind this rigorous connection and presents examples highlighting the utility of
quantitative molecular orbital theory in modern quantum chemistry.
Our strategy is to follow a specific observable and to define orbitals from the perspective
of how they relate to the chosen observable. This means that different observables give
rise to different orbitals. For example, Dyson orbitals are related to photodetachment and
photoionization cross sections probed by photoelectron spectroscopy (or to electron-impact
ionization cross sections), natural transition orbitals are related to linear spectroscopies (UV-
VIS, XAS, etc), natural (density) difference orbitals (giving rise to attachment-detachment
densities) are related to spectroscopies probing differences in states’ properties (e.g., Stark-
effect spectroscopies), and so on. Thus, we abandon the idea of a single definition of the
best molecular orbitals—in the author’s opinion, they do not exist. I also do not consider
orbital definitions that cannot be related to potentially measurable quantities—in this re-
spect, I share the sentiment that science should concern itself with observables. Different
experiments probe different types of orbitals—hence, what you see depends on the tool you
![Page 5: From orbitals to observables and backiopenshell.usc.edu/pubs/pdf/jcp_perspective.pdfQuantum mechanics explains chemical and physical phenomena by using abstract math-ematical constructs,](https://reader034.vdocuments.us/reader034/viewer/2022042406/5f1fa754210ddc366d060e82/html5/thumbnails/5.jpg)
5
use to interrogate your system. This lack of a single definition of the true orbitals is not a
flaw of the theory, rather, it is fully in accord with the quantum mechanical world, in which
the probe and the system that is probed are interconnected.
II. TRUE ORBITALS FOR EVERY OCCASION
A. Dyson orbitals
As the first example of experimentally defined orbitals, let us consider a photoioniza-
tion/photodetachment experiment. Within dipole approximation, the probability of ejecting
an electron is given by the photoelectron matrix element9:
Dk = u〈ΨNI |µ|ΨN−1
F Ψelk 〉, (2)
where µ is the dipole-moment operator, u is a unit vector along the polarization of ionizing
radiation, ΨNI and ΨN−1
F denote the initial and the ionized target state, and Ψelk is the
wavefunction of the ejected electron. Both ΨNI and ΨN−1
F are many-body wavefunctions,
but their explicit knowledge is not required for computing matrix elements of one-electron
operators, such as, for example, the dipole moment
µ = −N∑
i=1
ri. (3)
Eq. (2) is equivalent to10:
Dk = −u〈φd(r)|r|Ψelk (r)〉. (4)
Here φd(r) is a one-electron function, which contains all the necessary information about
the molecular system before and after it is ionized,
φd(1) =√N
∫ΨNI (1, . . . , n)ΨN−1
F (2, . . . , n)d2 . . . dn =∑
p
〈ΨNI |p†|ΨN−1
F 〉φp(1), (5)
where p† is the creation operator corresponding to orbital φp from the set of orthonormal
spin-orbitals used to construct the many-electron basis functions (Slater determinants used
to expand Ψ). In the derivation of this result11–14 we relied on the indistinguishability of
the electrons, which allowed us to carry out the integration over N electrons in two steps,
thus reducing the many-body integral in Eq. (2) to integration over the coordinates of just
![Page 6: From orbitals to observables and backiopenshell.usc.edu/pubs/pdf/jcp_perspective.pdfQuantum mechanics explains chemical and physical phenomena by using abstract math-ematical constructs,](https://reader034.vdocuments.us/reader034/viewer/2022042406/5f1fa754210ddc366d060e82/html5/thumbnails/6.jpg)
6
one electron. The only approximation here is the assumption that Ψelk is orthogonal to
the molecular core, which is usually justified by the large size of the free-electron state (this
assumption can be lifted, giving rise to a slightly bulkier expression, with overlap corrections
to the main term).
This function, φd(r), defined by Eq. (5), is called a Dyson orbital (or a generalized overlap
amplitude)11,14–20. Dyson orbitals can be defined for any type of wavefunction, including the
exact one—thus, they do not rely on the pseudo-non-interacting electron model21. They
simply tell you what the difference between the N and N -1 electron system is. Thus, they
can be interpreted as the initial state of the ejected electron.
Dyson orbitals also appear in the electron-impact ionization (e, 2e) cross sections22,23.
Electron-impact ionization is an example of electron-momentum spectroscopy, formulated
in the momentum space. The (e, 2e) cross section is proportional to the energy-momentum
spectral function:
S(Ei,p) =∑
av
|φdi (p)|2, (6)
where∑
av denotes averaging over the initial and final state degeneracies and φdi (p) is the
Dyson orbital corresponding to ionized state i (with ionization energy Ei) in the momentum
space:
φdi (p) =1
(2π)3/2
∫e−iprφdi (r)dr. (7)
The Dyson orbital in the coordinate space can be obtained by the reverse Fourier transform
of φdi (p).
Because Dyson orbitals are directly related to the
photoionization/photodetachment/electron-impact ionization cross sections, they can
be reconstructed from the experimental data and are, therefore, observable quantities.
Although the process of the reconstruction is not at all trivial (more on this below), here
we merely point out that experimental imaging of Dyson orbitals is, in principle, possible.
Alternatively, we can compute experimental observables from a given Dyson orbital and
judge the quality of the computed wavefunctions based on how well the theoretical and
experimental results agree. An example in Fig. 2 illustrates this point. Absolute pho-
toionization cross sections are very sensitive to the quality of Dyson orbitals. In a detailed
benchmark study14, we compared experimental absolute cross sections of the photoionization
of small molecules with the theoretical values based on the Dyson orbitals computed from
![Page 7: From orbitals to observables and backiopenshell.usc.edu/pubs/pdf/jcp_perspective.pdfQuantum mechanics explains chemical and physical phenomena by using abstract math-ematical constructs,](https://reader034.vdocuments.us/reader034/viewer/2022042406/5f1fa754210ddc366d060e82/html5/thumbnails/7.jpg)
7
0.0
0.2
0.4
0.6
0.8
10.5 11.0 11.5 12.0 12.5
Theory
Tota
l Cro
ss-S
ectio
n (a
.u.)
Energy of Ionizing Radiation (eV)
FIG. 2: Absolute photoionization yield of formaldehyde. In this energy range, photoelectron is
removed from the in-plane lone pair orbital (shown on the graph). The experimental values of the
cross section (shown in atomic units) are well reproduced by calculation using correlated Dyson
orbitals14.
coupled-cluster (CC) and equation-of-motion coupled-cluster (EOM-CC) wavefunctions24–26.
Figure 2 shows such a comparison for formaldehyde. As one can see, theory and experiment
are in perfect agreement. Importantly, using lower-quality orbitals (e.g., Kohn–Sham or
Hartree–Fock) leads to noticeable discrepancies. Thus, the ability to reproduce the absolute
photoionization cross sections provides a measure of the closeness of the shape of the Dyson
orbital to the exact one (of course, the calculation also requires the photoelectron wavefunc-
tion, as per Eq. (4); hence, the quality of the computed cross section depends also on the
quality of Ψelk ).
However, although total cross sections are highly sensitive to the shape of the Dyson
orbital, we cannot really reconstruct the orbital from the total cross section, which is a
too-deeply integrated quantity. Indirect information about orbital shapes can be deduced
from vibrational progressions and Franck–Condon factors, but to really see the orbitals,
one needs more detailed experimental observables. Differential cross sections, which provide
angular-resolved photoelectron spectra, contain such information. For randomly oriented
molecules, the photoelectron angular distribution (PAD) is completely characterized by the
dipole anisotropy parameter β: β ∼0 corresponds to isotropic distribution of photoelectrons,
β ∼ 2 corresponds to the photoelectrons aligned parallel to the polarization of the ionizing
radiation, and so on. Patterns in the PADs can be rationalized from dipole selection rules.
![Page 8: From orbitals to observables and backiopenshell.usc.edu/pubs/pdf/jcp_perspective.pdfQuantum mechanics explains chemical and physical phenomena by using abstract math-ematical constructs,](https://reader034.vdocuments.us/reader034/viewer/2022042406/5f1fa754210ddc366d060e82/html5/thumbnails/8.jpg)
8
As illustrated in Fig. 3, ionization from an s-orbital gives rise to a p-wave (with β=2),
ionization of a p-orbital gives rise to interfering s- and d-waves, and so on.Photoelectron Spectroscopy and Imaging
referring the reader to ref. 16 for a description of the rest ofour instrumentation.
In the idealized case of a monoenergetic electron and in theabsence of an external field, the position of the photoelectronat a given time t (measured from the photodetachment event)will lie on the surface of a sphere of radius vt. The center of thesphere coincides with the photoelectron origin. From thequantum mechanical perspective, the probability distributionof the position of the electron on the surface of the sphereconforms to |c|2, where c is the photoelectron wavefunction.Only when the position is measured, can we talk about alocalized position of the electron. Suppose for a moment thatwe employ a spherical position-sensitive detector surroundingthe photoelectron origin. Measurement with this detector willlocate the electron at a particular point on the sphere. That is,the act of measurement collapses the delocalized probabilitydistribution on the single point of observed electron impact.Repetitions of the experiment under exactly the sameconditions generally yield different impact positions. A largenumber of such repetitions will eventually map out the |c|2
probability density on the surface of the detector.In reality a spherical detector is impractical for photo-
electron spectroscopic measurements. Instead, the inherentcharge of the electron allows the use of electric fields to collectall of the photoelectrons by accelerating them towards a planardetector located some distance away from the photodetachmentorigin (see Fig. 1). The detector usually consists of a pair ofmicrochannel plates acting as a position-sensitive electronmultiplier, equipped with a phosphor screen. Impacts appearingon the screen are accumulated using a digital camera.
For a point electron source and uniform electric fielddirected along the y axis, which is perpendicular to thedetector plane (xz), this accumulation would eventually yielda 2-D projection of the 3-D probability density onto thedetector. Furthermore, since the applied field is perpendicularto the detector face the image would also represent the velocitydistribution within the xz plane. For multiple electron sources(ions spread throughout the interaction volume), any imagerecorded using a uniform field directed along y will consist of aseries of overlapping velocity distributions, resulting in spatialblurring. The ‘‘velocity mapping’’ electrode arrangement firstintroduced by Eppink and Parker2 minimizes this problem.The three electrodes shown in Fig. 1 create an inhomogeneouselectric field. In classical terms we might think of this asfunneling electrons onto specific points on the detector,which depend only on the velocity vector, regardless of thestarting point. In quantum mechanical terms the part of theprobability density corresponding to a particular electronvelocity is mapped onto a specific point on the detector(the reader is referred to ref. 2 for a full description). Withinthis ‘‘velocity mapping’’ arrangement the final imagerepresents the momentum–space distribution of the electronsin the plane of the detector.2,17
The photoelectron distribution produced with a polarized(or even an unpolarized, but directional) laser beam is notnecessarily isotropic. It is these anisotropic characteristicsof photodetachment that enable one to gain insights intothe symmetry properties of the parent-ion electron orbitalsand the dynamics of the detachment process. For linear
polarization, the electric field vector (ep) defines a unique axisin space (z). The resulting 3-D velocity distribution of thephotoelectrons varies with respect to y, the angle between theelectron velocity vector and the z axis, but is cylindricallysymmetric about z. The cylindrical symmetry allows amathematical reconstruction of the nascent 3-D velocitydistribution from its 2-D projection onto the detector planeusing the inverse Abel transformation.
3. The photoelectron image
The following discussion is based on the photoelectron imagesshown in Fig. 2 and 3, corresponding to the photodetachmentof H! at 800 nm and I! at 267 nm, respectively. For H!, asimilar 796 nm result was reported by Reichle et al.18
The process of image acquisition makes a compellingillustration of the Born interpretation of the wavefunctionand also demonstrates the nature of measurement in quantummechanics. As discussed in the previous section, individualelectrons are observed as localized impacts on the detectorsurface (Fig. 2a). With limited statistics, the impact spotsappear randomly distributed, but after many experimentalcycles, a pattern emerges (Fig. 2b) reflecting the probabilitydensity distribution. The radial intensity distributions in these
Fig. 2 Photoelectron imaging of H! at 800 nm. (a) An image
accumulated for 7 laser pulses, showing B10 randomly distributed
electron impacts (seen as bright spots). (b) 140 experimental cycles.
(c) 1.1 " 105 experimental cycles. The image is a 2-D projection of the
3-D distribution, which can be recovered from the inverse Abel
transformation, shown in (d). The photoelectron spectrum (e) and the
angular distribution (f) are extracted from the inverse Abel transformed
image. The laser polarization is vertical in the figure plane.
This journal is #c The Royal Society of Chemistry 2009 Chem. Soc. Rev., 2009, 38, 2169–2177 | 2171
Publ
ished
on
12 M
ay 2
009.
Dow
nloa
ded
by U
nive
rsity
of S
outh
ern
Calif
orni
a on
13/
05/2
014
22:3
2:37
.
View Article Online
referring the reader to ref. 16 for a description of the rest ofour instrumentation.
In the idealized case of a monoenergetic electron and in theabsence of an external field, the position of the photoelectronat a given time t (measured from the photodetachment event)will lie on the surface of a sphere of radius vt. The center of thesphere coincides with the photoelectron origin. From thequantum mechanical perspective, the probability distributionof the position of the electron on the surface of the sphereconforms to |c|2, where c is the photoelectron wavefunction.Only when the position is measured, can we talk about alocalized position of the electron. Suppose for a moment thatwe employ a spherical position-sensitive detector surroundingthe photoelectron origin. Measurement with this detector willlocate the electron at a particular point on the sphere. That is,the act of measurement collapses the delocalized probabilitydistribution on the single point of observed electron impact.Repetitions of the experiment under exactly the sameconditions generally yield different impact positions. A largenumber of such repetitions will eventually map out the |c|2
probability density on the surface of the detector.In reality a spherical detector is impractical for photo-
electron spectroscopic measurements. Instead, the inherentcharge of the electron allows the use of electric fields to collectall of the photoelectrons by accelerating them towards a planardetector located some distance away from the photodetachmentorigin (see Fig. 1). The detector usually consists of a pair ofmicrochannel plates acting as a position-sensitive electronmultiplier, equipped with a phosphor screen. Impacts appearingon the screen are accumulated using a digital camera.
For a point electron source and uniform electric fielddirected along the y axis, which is perpendicular to thedetector plane (xz), this accumulation would eventually yielda 2-D projection of the 3-D probability density onto thedetector. Furthermore, since the applied field is perpendicularto the detector face the image would also represent the velocitydistribution within the xz plane. For multiple electron sources(ions spread throughout the interaction volume), any imagerecorded using a uniform field directed along y will consist of aseries of overlapping velocity distributions, resulting in spatialblurring. The ‘‘velocity mapping’’ electrode arrangement firstintroduced by Eppink and Parker2 minimizes this problem.The three electrodes shown in Fig. 1 create an inhomogeneouselectric field. In classical terms we might think of this asfunneling electrons onto specific points on the detector,which depend only on the velocity vector, regardless of thestarting point. In quantum mechanical terms the part of theprobability density corresponding to a particular electronvelocity is mapped onto a specific point on the detector(the reader is referred to ref. 2 for a full description). Withinthis ‘‘velocity mapping’’ arrangement the final imagerepresents the momentum–space distribution of the electronsin the plane of the detector.2,17
The photoelectron distribution produced with a polarized(or even an unpolarized, but directional) laser beam is notnecessarily isotropic. It is these anisotropic characteristicsof photodetachment that enable one to gain insights intothe symmetry properties of the parent-ion electron orbitalsand the dynamics of the detachment process. For linear
polarization, the electric field vector (ep) defines a unique axisin space (z). The resulting 3-D velocity distribution of thephotoelectrons varies with respect to y, the angle between theelectron velocity vector and the z axis, but is cylindricallysymmetric about z. The cylindrical symmetry allows amathematical reconstruction of the nascent 3-D velocitydistribution from its 2-D projection onto the detector planeusing the inverse Abel transformation.
3. The photoelectron image
The following discussion is based on the photoelectron imagesshown in Fig. 2 and 3, corresponding to the photodetachmentof H! at 800 nm and I! at 267 nm, respectively. For H!, asimilar 796 nm result was reported by Reichle et al.18
The process of image acquisition makes a compellingillustration of the Born interpretation of the wavefunctionand also demonstrates the nature of measurement in quantummechanics. As discussed in the previous section, individualelectrons are observed as localized impacts on the detectorsurface (Fig. 2a). With limited statistics, the impact spotsappear randomly distributed, but after many experimentalcycles, a pattern emerges (Fig. 2b) reflecting the probabilitydensity distribution. The radial intensity distributions in these
Fig. 2 Photoelectron imaging of H! at 800 nm. (a) An image
accumulated for 7 laser pulses, showing B10 randomly distributed
electron impacts (seen as bright spots). (b) 140 experimental cycles.
(c) 1.1 " 105 experimental cycles. The image is a 2-D projection of the
3-D distribution, which can be recovered from the inverse Abel
transformation, shown in (d). The photoelectron spectrum (e) and the
angular distribution (f) are extracted from the inverse Abel transformed
image. The laser polarization is vertical in the figure plane.
This journal is #c The Royal Society of Chemistry 2009 Chem. Soc. Rev., 2009, 38, 2169–2177 | 2171
Publ
ished
on
12 M
ay 2
009.
Dow
nloa
ded
by U
nive
rsity
of S
outh
ern
Calif
orni
a on
13/
05/2
014
22:3
2:37
.
View Article Online
εP
vθ
Mabbs, R.; Grumbling, E. R.; Pichugin, K.; Sanov, A. Chem. Soc. Rev., 2009, 38, 2169-2177
referring the reader to ref. 16 for a description of the rest ofour instrumentation.
In the idealized case of a monoenergetic electron and in theabsence of an external field, the position of the photoelectronat a given time t (measured from the photodetachment event)will lie on the surface of a sphere of radius vt. The center of thesphere coincides with the photoelectron origin. From thequantum mechanical perspective, the probability distributionof the position of the electron on the surface of the sphereconforms to |c|2, where c is the photoelectron wavefunction.Only when the position is measured, can we talk about alocalized position of the electron. Suppose for a moment thatwe employ a spherical position-sensitive detector surroundingthe photoelectron origin. Measurement with this detector willlocate the electron at a particular point on the sphere. That is,the act of measurement collapses the delocalized probabilitydistribution on the single point of observed electron impact.Repetitions of the experiment under exactly the sameconditions generally yield different impact positions. A largenumber of such repetitions will eventually map out the |c|2
probability density on the surface of the detector.In reality a spherical detector is impractical for photo-
electron spectroscopic measurements. Instead, the inherentcharge of the electron allows the use of electric fields to collectall of the photoelectrons by accelerating them towards a planardetector located some distance away from the photodetachmentorigin (see Fig. 1). The detector usually consists of a pair ofmicrochannel plates acting as a position-sensitive electronmultiplier, equipped with a phosphor screen. Impacts appearingon the screen are accumulated using a digital camera.
For a point electron source and uniform electric fielddirected along the y axis, which is perpendicular to thedetector plane (xz), this accumulation would eventually yielda 2-D projection of the 3-D probability density onto thedetector. Furthermore, since the applied field is perpendicularto the detector face the image would also represent the velocitydistribution within the xz plane. For multiple electron sources(ions spread throughout the interaction volume), any imagerecorded using a uniform field directed along y will consist of aseries of overlapping velocity distributions, resulting in spatialblurring. The ‘‘velocity mapping’’ electrode arrangement firstintroduced by Eppink and Parker2 minimizes this problem.The three electrodes shown in Fig. 1 create an inhomogeneouselectric field. In classical terms we might think of this asfunneling electrons onto specific points on the detector,which depend only on the velocity vector, regardless of thestarting point. In quantum mechanical terms the part of theprobability density corresponding to a particular electronvelocity is mapped onto a specific point on the detector(the reader is referred to ref. 2 for a full description). Withinthis ‘‘velocity mapping’’ arrangement the final imagerepresents the momentum–space distribution of the electronsin the plane of the detector.2,17
The photoelectron distribution produced with a polarized(or even an unpolarized, but directional) laser beam is notnecessarily isotropic. It is these anisotropic characteristicsof photodetachment that enable one to gain insights intothe symmetry properties of the parent-ion electron orbitalsand the dynamics of the detachment process. For linear
polarization, the electric field vector (ep) defines a unique axisin space (z). The resulting 3-D velocity distribution of thephotoelectrons varies with respect to y, the angle between theelectron velocity vector and the z axis, but is cylindricallysymmetric about z. The cylindrical symmetry allows amathematical reconstruction of the nascent 3-D velocitydistribution from its 2-D projection onto the detector planeusing the inverse Abel transformation.
3. The photoelectron image
The following discussion is based on the photoelectron imagesshown in Fig. 2 and 3, corresponding to the photodetachmentof H! at 800 nm and I! at 267 nm, respectively. For H!, asimilar 796 nm result was reported by Reichle et al.18
The process of image acquisition makes a compellingillustration of the Born interpretation of the wavefunctionand also demonstrates the nature of measurement in quantummechanics. As discussed in the previous section, individualelectrons are observed as localized impacts on the detectorsurface (Fig. 2a). With limited statistics, the impact spotsappear randomly distributed, but after many experimentalcycles, a pattern emerges (Fig. 2b) reflecting the probabilitydensity distribution. The radial intensity distributions in these
Fig. 2 Photoelectron imaging of H! at 800 nm. (a) An image
accumulated for 7 laser pulses, showing B10 randomly distributed
electron impacts (seen as bright spots). (b) 140 experimental cycles.
(c) 1.1 " 105 experimental cycles. The image is a 2-D projection of the
3-D distribution, which can be recovered from the inverse Abel
transformation, shown in (d). The photoelectron spectrum (e) and the
angular distribution (f) are extracted from the inverse Abel transformed
image. The laser polarization is vertical in the figure plane.
This journal is #c The Royal Society of Chemistry 2009 Chem. Soc. Rev., 2009, 38, 2169–2177 | 2171
Publ
ishe
d on
12
May
200
9. D
ownl
oade
d by
Uni
vers
ity o
f Sou
ther
n C
alifo
rnia
on
13/0
5/20
14 2
2:32
:37.
View Article Online
H-
I-
vθ
images show sharp maxima, while the angular distributionsare distinctly anisotropic with respect to the laser polarizationdirection. We will use these results to illustrate the basics ofimage analysis and interpretation.
3.1 The photoelectron spectrum
In the Abel-inverted image, the electron speed is proportionalto the distance r from the center. Integration over y at eachparticular r yields the radial distribution, or the photo-electron momentum spectrum, which can be converted to theconventional photoelectron energy spectrum via Jacobiantransformation. The spectra of H! and I! in Fig. 2e and 3care shown as functions of electron kinetic energy (eKE).
The peaks in a photoelectron spectrum correspond toelectron ejection accompanied by the population of specificenergy levels of the residual neutral. The single peak in thephotoelectron spectrum of H! is due to the removal of anelectron from the fully populated 1s orbital yielding the 2S1/2ground state of the neutral atom.
For I!, two spin–orbit states, significantly different inenergy, are accessed upon removal of a p-electron using267 nm radiation. The higher-eKE peak corresponds tophotoelectrons produced in coincidence with the ground(2P3/2) electronic state of the iodine atom, while the lowerpeak corresponds to the excited state (2P1/2). The 0.94 eVspacing between the peaks reflects the magnitude of thespin–orbit interaction in the iodine atom. The relativeintegrated peak intensities are proportional to the transitionprobabilities, including the degeneracies of the 2P3/2 and2P1/2 states of iodine.19 The finite peak widths in the energydomain reflect experimental broadening factors, such as laserlinewidth, parent-ion velocity spread, etc.16
3.2 The photoelectron angular distribution
The photoelectron angular distribution (PAD) is obtained byintegrating the intensity of the Abel-inverted image over achosen spectral (i.e. radial) range. By inspection, the H! PAD(Fig. 2f) has a ‘‘parallel’’ character, peaking at y = 0 and p,while the I! PADs (Fig. 3d) for both detachment channelspeak at y = p/2, corresponding to ‘‘perpendicular’’ photo-detachment transitions.The PADs in one-photon photodetachment with linearly
polarized light are generally described by the function:20
I(y) = (s/4p) [1 + bP2(cosy)], (2)
where the proportionality constant s corresponds to the totalphotodetachment cross-section and P2(cosy) " 1/2(3cos2y ! 1)is the second-order Legendre polynomial. The angulardependence is completely defined by b, the anisotropy para-meter, which can be determined by fitting eqn (2) to theexperimental PAD. The b values range from !1 to +2, withnegative and positive values corresponding to perpendicularand parallel polarization, respectively. The b value observed inH! photodetachment is very close to the +2 limit of a purelyparallel transition. For I! photodetachment, however, therings in Fig. 3(a) correspond to b E !0.6.
4. Interpreting photoelectron images
4.1 Relating photoelectron angular distributions to parentorbitals: the H! example
In a pulsed laser experiment the electron is described by asuperposition of waves spanning a finite energy range. In thelaboratory frame, this wavepacket is described as:
C(r,y,f,t) =Ra(k)Ck (r,y,f,t)dk (3)
where a(k) is a shape function related to the laser lineshape.Assuming a central, i.e. spherically symmetric, potentialCk(r,y,f,t) can be separated into radial and angular components,
Ck(r,y,f,t) = Rk(r,t)Y(y,f) (4)
The time dependence is carried in the radial function and thecorresponding electron probability density can be likened to apattern on a spherical balloon. As the ‘‘balloon’’ is inflated,the pattern expands, but its angular dependence remainsunchanged.The connection between a photoelectron image and the
parent orbital can be demonstrated using three different(yet closely inter-related) approaches based on: (1) angularmomentum conservation; (2) transition dipole momentoperations; (3) general symmetry considerations. In thefollowing, we adopt the one-electron picture of photo-detachment and discuss the uniquely simple case of H! toillustrate the key ideas in photoelectron image interpretation.
4.1.1 Conservation of angular momentum. The selectionrule for one-photon, one-electron atomic transitions isDc = #1, where c is the orbital angular momentum quantumnumber. In detachment from the hydrogen 1s orbital (ci = 0),the free-electron wave is characterized by a single c value of 1,corresponding to a p wave.
Fig. 3 Photoelectron imaging of I! at 267 nm. (a) The photoelectron
image. (b) Inverse Abel transformation. The two distinct rings corres-
pond to the different peaks in the photoelectron spectrum (c). The
photoelectron angular distribution (d) corresponding to the 2P3/2
transition peaks at p/2, in contrast to the H! detachment in Fig. 2
and contrary to classical expectations.
2172 | Chem. Soc. Rev., 2009, 38, 2169–2177 This journal is $c The Royal Society of Chemistry 2009
Publ
ished
on
12 M
ay 2
009.
Dow
nloa
ded
by U
nive
rsity
of S
outh
ern
Calif
orni
a on
13/
05/2
014
22:3
2:37
.
View Article Online
images show sharp maxima, while the angular distributionsare distinctly anisotropic with respect to the laser polarizationdirection. We will use these results to illustrate the basics ofimage analysis and interpretation.
3.1 The photoelectron spectrum
In the Abel-inverted image, the electron speed is proportionalto the distance r from the center. Integration over y at eachparticular r yields the radial distribution, or the photo-electron momentum spectrum, which can be converted to theconventional photoelectron energy spectrum via Jacobiantransformation. The spectra of H! and I! in Fig. 2e and 3care shown as functions of electron kinetic energy (eKE).
The peaks in a photoelectron spectrum correspond toelectron ejection accompanied by the population of specificenergy levels of the residual neutral. The single peak in thephotoelectron spectrum of H! is due to the removal of anelectron from the fully populated 1s orbital yielding the 2S1/2ground state of the neutral atom.
For I!, two spin–orbit states, significantly different inenergy, are accessed upon removal of a p-electron using267 nm radiation. The higher-eKE peak corresponds tophotoelectrons produced in coincidence with the ground(2P3/2) electronic state of the iodine atom, while the lowerpeak corresponds to the excited state (2P1/2). The 0.94 eVspacing between the peaks reflects the magnitude of thespin–orbit interaction in the iodine atom. The relativeintegrated peak intensities are proportional to the transitionprobabilities, including the degeneracies of the 2P3/2 and2P1/2 states of iodine.19 The finite peak widths in the energydomain reflect experimental broadening factors, such as laserlinewidth, parent-ion velocity spread, etc.16
3.2 The photoelectron angular distribution
The photoelectron angular distribution (PAD) is obtained byintegrating the intensity of the Abel-inverted image over achosen spectral (i.e. radial) range. By inspection, the H! PAD(Fig. 2f) has a ‘‘parallel’’ character, peaking at y = 0 and p,while the I! PADs (Fig. 3d) for both detachment channelspeak at y = p/2, corresponding to ‘‘perpendicular’’ photo-detachment transitions.The PADs in one-photon photodetachment with linearly
polarized light are generally described by the function:20
I(y) = (s/4p) [1 + bP2(cosy)], (2)
where the proportionality constant s corresponds to the totalphotodetachment cross-section and P2(cosy) " 1/2(3cos2y ! 1)is the second-order Legendre polynomial. The angulardependence is completely defined by b, the anisotropy para-meter, which can be determined by fitting eqn (2) to theexperimental PAD. The b values range from !1 to +2, withnegative and positive values corresponding to perpendicularand parallel polarization, respectively. The b value observed inH! photodetachment is very close to the +2 limit of a purelyparallel transition. For I! photodetachment, however, therings in Fig. 3(a) correspond to b E !0.6.
4. Interpreting photoelectron images
4.1 Relating photoelectron angular distributions to parentorbitals: the H! example
In a pulsed laser experiment the electron is described by asuperposition of waves spanning a finite energy range. In thelaboratory frame, this wavepacket is described as:
C(r,y,f,t) =Ra(k)Ck (r,y,f,t)dk (3)
where a(k) is a shape function related to the laser lineshape.Assuming a central, i.e. spherically symmetric, potentialCk(r,y,f,t) can be separated into radial and angular components,
Ck(r,y,f,t) = Rk(r,t)Y(y,f) (4)
The time dependence is carried in the radial function and thecorresponding electron probability density can be likened to apattern on a spherical balloon. As the ‘‘balloon’’ is inflated,the pattern expands, but its angular dependence remainsunchanged.The connection between a photoelectron image and the
parent orbital can be demonstrated using three different(yet closely inter-related) approaches based on: (1) angularmomentum conservation; (2) transition dipole momentoperations; (3) general symmetry considerations. In thefollowing, we adopt the one-electron picture of photo-detachment and discuss the uniquely simple case of H! toillustrate the key ideas in photoelectron image interpretation.
4.1.1 Conservation of angular momentum. The selectionrule for one-photon, one-electron atomic transitions isDc = #1, where c is the orbital angular momentum quantumnumber. In detachment from the hydrogen 1s orbital (ci = 0),the free-electron wave is characterized by a single c value of 1,corresponding to a p wave.
Fig. 3 Photoelectron imaging of I! at 267 nm. (a) The photoelectron
image. (b) Inverse Abel transformation. The two distinct rings corres-
pond to the different peaks in the photoelectron spectrum (c). The
photoelectron angular distribution (d) corresponding to the 2P3/2
transition peaks at p/2, in contrast to the H! detachment in Fig. 2
and contrary to classical expectations.
2172 | Chem. Soc. Rev., 2009, 38, 2169–2177 This journal is $c The Royal Society of Chemistry 2009
Publ
ished
on
12 M
ay 2
009.
Dow
nloa
ded
by U
nive
rsity
of S
outh
ern
Calif
orni
a on
13/
05/2
014
22:3
2:37
.
View Article OnlinePhotoelectron Spectroscopy and Imaging
referring the reader to ref. 16 for a description of the rest ofour instrumentation.
In the idealized case of a monoenergetic electron and in theabsence of an external field, the position of the photoelectronat a given time t (measured from the photodetachment event)will lie on the surface of a sphere of radius vt. The center of thesphere coincides with the photoelectron origin. From thequantum mechanical perspective, the probability distributionof the position of the electron on the surface of the sphereconforms to |c|2, where c is the photoelectron wavefunction.Only when the position is measured, can we talk about alocalized position of the electron. Suppose for a moment thatwe employ a spherical position-sensitive detector surroundingthe photoelectron origin. Measurement with this detector willlocate the electron at a particular point on the sphere. That is,the act of measurement collapses the delocalized probabilitydistribution on the single point of observed electron impact.Repetitions of the experiment under exactly the sameconditions generally yield different impact positions. A largenumber of such repetitions will eventually map out the |c|2
probability density on the surface of the detector.In reality a spherical detector is impractical for photo-
electron spectroscopic measurements. Instead, the inherentcharge of the electron allows the use of electric fields to collectall of the photoelectrons by accelerating them towards a planardetector located some distance away from the photodetachmentorigin (see Fig. 1). The detector usually consists of a pair ofmicrochannel plates acting as a position-sensitive electronmultiplier, equipped with a phosphor screen. Impacts appearingon the screen are accumulated using a digital camera.
For a point electron source and uniform electric fielddirected along the y axis, which is perpendicular to thedetector plane (xz), this accumulation would eventually yielda 2-D projection of the 3-D probability density onto thedetector. Furthermore, since the applied field is perpendicularto the detector face the image would also represent the velocitydistribution within the xz plane. For multiple electron sources(ions spread throughout the interaction volume), any imagerecorded using a uniform field directed along y will consist of aseries of overlapping velocity distributions, resulting in spatialblurring. The ‘‘velocity mapping’’ electrode arrangement firstintroduced by Eppink and Parker2 minimizes this problem.The three electrodes shown in Fig. 1 create an inhomogeneouselectric field. In classical terms we might think of this asfunneling electrons onto specific points on the detector,which depend only on the velocity vector, regardless of thestarting point. In quantum mechanical terms the part of theprobability density corresponding to a particular electronvelocity is mapped onto a specific point on the detector(the reader is referred to ref. 2 for a full description). Withinthis ‘‘velocity mapping’’ arrangement the final imagerepresents the momentum–space distribution of the electronsin the plane of the detector.2,17
The photoelectron distribution produced with a polarized(or even an unpolarized, but directional) laser beam is notnecessarily isotropic. It is these anisotropic characteristicsof photodetachment that enable one to gain insights intothe symmetry properties of the parent-ion electron orbitalsand the dynamics of the detachment process. For linear
polarization, the electric field vector (ep) defines a unique axisin space (z). The resulting 3-D velocity distribution of thephotoelectrons varies with respect to y, the angle between theelectron velocity vector and the z axis, but is cylindricallysymmetric about z. The cylindrical symmetry allows amathematical reconstruction of the nascent 3-D velocitydistribution from its 2-D projection onto the detector planeusing the inverse Abel transformation.
3. The photoelectron image
The following discussion is based on the photoelectron imagesshown in Fig. 2 and 3, corresponding to the photodetachmentof H! at 800 nm and I! at 267 nm, respectively. For H!, asimilar 796 nm result was reported by Reichle et al.18
The process of image acquisition makes a compellingillustration of the Born interpretation of the wavefunctionand also demonstrates the nature of measurement in quantummechanics. As discussed in the previous section, individualelectrons are observed as localized impacts on the detectorsurface (Fig. 2a). With limited statistics, the impact spotsappear randomly distributed, but after many experimentalcycles, a pattern emerges (Fig. 2b) reflecting the probabilitydensity distribution. The radial intensity distributions in these
Fig. 2 Photoelectron imaging of H! at 800 nm. (a) An image
accumulated for 7 laser pulses, showing B10 randomly distributed
electron impacts (seen as bright spots). (b) 140 experimental cycles.
(c) 1.1 " 105 experimental cycles. The image is a 2-D projection of the
3-D distribution, which can be recovered from the inverse Abel
transformation, shown in (d). The photoelectron spectrum (e) and the
angular distribution (f) are extracted from the inverse Abel transformed
image. The laser polarization is vertical in the figure plane.
This journal is #c The Royal Society of Chemistry 2009 Chem. Soc. Rev., 2009, 38, 2169–2177 | 2171
Publ
ished
on
12 M
ay 2
009.
Dow
nloa
ded
by U
nive
rsity
of S
outh
ern
Calif
orni
a on
13/
05/2
014
22:3
2:37
.
View Article Online
referring the reader to ref. 16 for a description of the rest ofour instrumentation.
In the idealized case of a monoenergetic electron and in theabsence of an external field, the position of the photoelectronat a given time t (measured from the photodetachment event)will lie on the surface of a sphere of radius vt. The center of thesphere coincides with the photoelectron origin. From thequantum mechanical perspective, the probability distributionof the position of the electron on the surface of the sphereconforms to |c|2, where c is the photoelectron wavefunction.Only when the position is measured, can we talk about alocalized position of the electron. Suppose for a moment thatwe employ a spherical position-sensitive detector surroundingthe photoelectron origin. Measurement with this detector willlocate the electron at a particular point on the sphere. That is,the act of measurement collapses the delocalized probabilitydistribution on the single point of observed electron impact.Repetitions of the experiment under exactly the sameconditions generally yield different impact positions. A largenumber of such repetitions will eventually map out the |c|2
probability density on the surface of the detector.In reality a spherical detector is impractical for photo-
electron spectroscopic measurements. Instead, the inherentcharge of the electron allows the use of electric fields to collectall of the photoelectrons by accelerating them towards a planardetector located some distance away from the photodetachmentorigin (see Fig. 1). The detector usually consists of a pair ofmicrochannel plates acting as a position-sensitive electronmultiplier, equipped with a phosphor screen. Impacts appearingon the screen are accumulated using a digital camera.
For a point electron source and uniform electric fielddirected along the y axis, which is perpendicular to thedetector plane (xz), this accumulation would eventually yielda 2-D projection of the 3-D probability density onto thedetector. Furthermore, since the applied field is perpendicularto the detector face the image would also represent the velocitydistribution within the xz plane. For multiple electron sources(ions spread throughout the interaction volume), any imagerecorded using a uniform field directed along y will consist of aseries of overlapping velocity distributions, resulting in spatialblurring. The ‘‘velocity mapping’’ electrode arrangement firstintroduced by Eppink and Parker2 minimizes this problem.The three electrodes shown in Fig. 1 create an inhomogeneouselectric field. In classical terms we might think of this asfunneling electrons onto specific points on the detector,which depend only on the velocity vector, regardless of thestarting point. In quantum mechanical terms the part of theprobability density corresponding to a particular electronvelocity is mapped onto a specific point on the detector(the reader is referred to ref. 2 for a full description). Withinthis ‘‘velocity mapping’’ arrangement the final imagerepresents the momentum–space distribution of the electronsin the plane of the detector.2,17
The photoelectron distribution produced with a polarized(or even an unpolarized, but directional) laser beam is notnecessarily isotropic. It is these anisotropic characteristicsof photodetachment that enable one to gain insights intothe symmetry properties of the parent-ion electron orbitalsand the dynamics of the detachment process. For linear
polarization, the electric field vector (ep) defines a unique axisin space (z). The resulting 3-D velocity distribution of thephotoelectrons varies with respect to y, the angle between theelectron velocity vector and the z axis, but is cylindricallysymmetric about z. The cylindrical symmetry allows amathematical reconstruction of the nascent 3-D velocitydistribution from its 2-D projection onto the detector planeusing the inverse Abel transformation.
3. The photoelectron image
The following discussion is based on the photoelectron imagesshown in Fig. 2 and 3, corresponding to the photodetachmentof H! at 800 nm and I! at 267 nm, respectively. For H!, asimilar 796 nm result was reported by Reichle et al.18
The process of image acquisition makes a compellingillustration of the Born interpretation of the wavefunctionand also demonstrates the nature of measurement in quantummechanics. As discussed in the previous section, individualelectrons are observed as localized impacts on the detectorsurface (Fig. 2a). With limited statistics, the impact spotsappear randomly distributed, but after many experimentalcycles, a pattern emerges (Fig. 2b) reflecting the probabilitydensity distribution. The radial intensity distributions in these
Fig. 2 Photoelectron imaging of H! at 800 nm. (a) An image
accumulated for 7 laser pulses, showing B10 randomly distributed
electron impacts (seen as bright spots). (b) 140 experimental cycles.
(c) 1.1 " 105 experimental cycles. The image is a 2-D projection of the
3-D distribution, which can be recovered from the inverse Abel
transformation, shown in (d). The photoelectron spectrum (e) and the
angular distribution (f) are extracted from the inverse Abel transformed
image. The laser polarization is vertical in the figure plane.
This journal is #c The Royal Society of Chemistry 2009 Chem. Soc. Rev., 2009, 38, 2169–2177 | 2171
Publ
ished
on
12 M
ay 2
009.
Dow
nloa
ded
by U
nive
rsity
of S
outh
ern
Calif
orni
a on
13/
05/2
014
22:3
2:37
.
View Article Online
εP
vθ
Mabbs, R.; Grumbling, E. R.; Pichugin, K.; Sanov, A. Chem. Soc. Rev., 2009, 38, 2169-2177
referring the reader to ref. 16 for a description of the rest ofour instrumentation.
In the idealized case of a monoenergetic electron and in theabsence of an external field, the position of the photoelectronat a given time t (measured from the photodetachment event)will lie on the surface of a sphere of radius vt. The center of thesphere coincides with the photoelectron origin. From thequantum mechanical perspective, the probability distributionof the position of the electron on the surface of the sphereconforms to |c|2, where c is the photoelectron wavefunction.Only when the position is measured, can we talk about alocalized position of the electron. Suppose for a moment thatwe employ a spherical position-sensitive detector surroundingthe photoelectron origin. Measurement with this detector willlocate the electron at a particular point on the sphere. That is,the act of measurement collapses the delocalized probabilitydistribution on the single point of observed electron impact.Repetitions of the experiment under exactly the sameconditions generally yield different impact positions. A largenumber of such repetitions will eventually map out the |c|2
probability density on the surface of the detector.In reality a spherical detector is impractical for photo-
electron spectroscopic measurements. Instead, the inherentcharge of the electron allows the use of electric fields to collectall of the photoelectrons by accelerating them towards a planardetector located some distance away from the photodetachmentorigin (see Fig. 1). The detector usually consists of a pair ofmicrochannel plates acting as a position-sensitive electronmultiplier, equipped with a phosphor screen. Impacts appearingon the screen are accumulated using a digital camera.
For a point electron source and uniform electric fielddirected along the y axis, which is perpendicular to thedetector plane (xz), this accumulation would eventually yielda 2-D projection of the 3-D probability density onto thedetector. Furthermore, since the applied field is perpendicularto the detector face the image would also represent the velocitydistribution within the xz plane. For multiple electron sources(ions spread throughout the interaction volume), any imagerecorded using a uniform field directed along y will consist of aseries of overlapping velocity distributions, resulting in spatialblurring. The ‘‘velocity mapping’’ electrode arrangement firstintroduced by Eppink and Parker2 minimizes this problem.The three electrodes shown in Fig. 1 create an inhomogeneouselectric field. In classical terms we might think of this asfunneling electrons onto specific points on the detector,which depend only on the velocity vector, regardless of thestarting point. In quantum mechanical terms the part of theprobability density corresponding to a particular electronvelocity is mapped onto a specific point on the detector(the reader is referred to ref. 2 for a full description). Withinthis ‘‘velocity mapping’’ arrangement the final imagerepresents the momentum–space distribution of the electronsin the plane of the detector.2,17
The photoelectron distribution produced with a polarized(or even an unpolarized, but directional) laser beam is notnecessarily isotropic. It is these anisotropic characteristicsof photodetachment that enable one to gain insights intothe symmetry properties of the parent-ion electron orbitalsand the dynamics of the detachment process. For linear
polarization, the electric field vector (ep) defines a unique axisin space (z). The resulting 3-D velocity distribution of thephotoelectrons varies with respect to y, the angle between theelectron velocity vector and the z axis, but is cylindricallysymmetric about z. The cylindrical symmetry allows amathematical reconstruction of the nascent 3-D velocitydistribution from its 2-D projection onto the detector planeusing the inverse Abel transformation.
3. The photoelectron image
The following discussion is based on the photoelectron imagesshown in Fig. 2 and 3, corresponding to the photodetachmentof H! at 800 nm and I! at 267 nm, respectively. For H!, asimilar 796 nm result was reported by Reichle et al.18
The process of image acquisition makes a compellingillustration of the Born interpretation of the wavefunctionand also demonstrates the nature of measurement in quantummechanics. As discussed in the previous section, individualelectrons are observed as localized impacts on the detectorsurface (Fig. 2a). With limited statistics, the impact spotsappear randomly distributed, but after many experimentalcycles, a pattern emerges (Fig. 2b) reflecting the probabilitydensity distribution. The radial intensity distributions in these
Fig. 2 Photoelectron imaging of H! at 800 nm. (a) An image
accumulated for 7 laser pulses, showing B10 randomly distributed
electron impacts (seen as bright spots). (b) 140 experimental cycles.
(c) 1.1 " 105 experimental cycles. The image is a 2-D projection of the
3-D distribution, which can be recovered from the inverse Abel
transformation, shown in (d). The photoelectron spectrum (e) and the
angular distribution (f) are extracted from the inverse Abel transformed
image. The laser polarization is vertical in the figure plane.
This journal is #c The Royal Society of Chemistry 2009 Chem. Soc. Rev., 2009, 38, 2169–2177 | 2171
Publ
ishe
d on
12
May
200
9. D
ownl
oade
d by
Uni
vers
ity o
f Sou
ther
n C
alifo
rnia
on
13/0
5/20
14 2
2:32
:37.
View Article Online
H-
I-
vθ
images show sharp maxima, while the angular distributionsare distinctly anisotropic with respect to the laser polarizationdirection. We will use these results to illustrate the basics ofimage analysis and interpretation.
3.1 The photoelectron spectrum
In the Abel-inverted image, the electron speed is proportionalto the distance r from the center. Integration over y at eachparticular r yields the radial distribution, or the photo-electron momentum spectrum, which can be converted to theconventional photoelectron energy spectrum via Jacobiantransformation. The spectra of H! and I! in Fig. 2e and 3care shown as functions of electron kinetic energy (eKE).
The peaks in a photoelectron spectrum correspond toelectron ejection accompanied by the population of specificenergy levels of the residual neutral. The single peak in thephotoelectron spectrum of H! is due to the removal of anelectron from the fully populated 1s orbital yielding the 2S1/2ground state of the neutral atom.
For I!, two spin–orbit states, significantly different inenergy, are accessed upon removal of a p-electron using267 nm radiation. The higher-eKE peak corresponds tophotoelectrons produced in coincidence with the ground(2P3/2) electronic state of the iodine atom, while the lowerpeak corresponds to the excited state (2P1/2). The 0.94 eVspacing between the peaks reflects the magnitude of thespin–orbit interaction in the iodine atom. The relativeintegrated peak intensities are proportional to the transitionprobabilities, including the degeneracies of the 2P3/2 and2P1/2 states of iodine.19 The finite peak widths in the energydomain reflect experimental broadening factors, such as laserlinewidth, parent-ion velocity spread, etc.16
3.2 The photoelectron angular distribution
The photoelectron angular distribution (PAD) is obtained byintegrating the intensity of the Abel-inverted image over achosen spectral (i.e. radial) range. By inspection, the H! PAD(Fig. 2f) has a ‘‘parallel’’ character, peaking at y = 0 and p,while the I! PADs (Fig. 3d) for both detachment channelspeak at y = p/2, corresponding to ‘‘perpendicular’’ photo-detachment transitions.The PADs in one-photon photodetachment with linearly
polarized light are generally described by the function:20
I(y) = (s/4p) [1 + bP2(cosy)], (2)
where the proportionality constant s corresponds to the totalphotodetachment cross-section and P2(cosy) " 1/2(3cos2y ! 1)is the second-order Legendre polynomial. The angulardependence is completely defined by b, the anisotropy para-meter, which can be determined by fitting eqn (2) to theexperimental PAD. The b values range from !1 to +2, withnegative and positive values corresponding to perpendicularand parallel polarization, respectively. The b value observed inH! photodetachment is very close to the +2 limit of a purelyparallel transition. For I! photodetachment, however, therings in Fig. 3(a) correspond to b E !0.6.
4. Interpreting photoelectron images
4.1 Relating photoelectron angular distributions to parentorbitals: the H! example
In a pulsed laser experiment the electron is described by asuperposition of waves spanning a finite energy range. In thelaboratory frame, this wavepacket is described as:
C(r,y,f,t) =Ra(k)Ck (r,y,f,t)dk (3)
where a(k) is a shape function related to the laser lineshape.Assuming a central, i.e. spherically symmetric, potentialCk(r,y,f,t) can be separated into radial and angular components,
Ck(r,y,f,t) = Rk(r,t)Y(y,f) (4)
The time dependence is carried in the radial function and thecorresponding electron probability density can be likened to apattern on a spherical balloon. As the ‘‘balloon’’ is inflated,the pattern expands, but its angular dependence remainsunchanged.The connection between a photoelectron image and the
parent orbital can be demonstrated using three different(yet closely inter-related) approaches based on: (1) angularmomentum conservation; (2) transition dipole momentoperations; (3) general symmetry considerations. In thefollowing, we adopt the one-electron picture of photo-detachment and discuss the uniquely simple case of H! toillustrate the key ideas in photoelectron image interpretation.
4.1.1 Conservation of angular momentum. The selectionrule for one-photon, one-electron atomic transitions isDc = #1, where c is the orbital angular momentum quantumnumber. In detachment from the hydrogen 1s orbital (ci = 0),the free-electron wave is characterized by a single c value of 1,corresponding to a p wave.
Fig. 3 Photoelectron imaging of I! at 267 nm. (a) The photoelectron
image. (b) Inverse Abel transformation. The two distinct rings corres-
pond to the different peaks in the photoelectron spectrum (c). The
photoelectron angular distribution (d) corresponding to the 2P3/2
transition peaks at p/2, in contrast to the H! detachment in Fig. 2
and contrary to classical expectations.
2172 | Chem. Soc. Rev., 2009, 38, 2169–2177 This journal is $c The Royal Society of Chemistry 2009
Publ
ished
on
12 M
ay 2
009.
Dow
nloa
ded
by U
nive
rsity
of S
outh
ern
Calif
orni
a on
13/
05/2
014
22:3
2:37
.
View Article Online
images show sharp maxima, while the angular distributionsare distinctly anisotropic with respect to the laser polarizationdirection. We will use these results to illustrate the basics ofimage analysis and interpretation.
3.1 The photoelectron spectrum
In the Abel-inverted image, the electron speed is proportionalto the distance r from the center. Integration over y at eachparticular r yields the radial distribution, or the photo-electron momentum spectrum, which can be converted to theconventional photoelectron energy spectrum via Jacobiantransformation. The spectra of H! and I! in Fig. 2e and 3care shown as functions of electron kinetic energy (eKE).
The peaks in a photoelectron spectrum correspond toelectron ejection accompanied by the population of specificenergy levels of the residual neutral. The single peak in thephotoelectron spectrum of H! is due to the removal of anelectron from the fully populated 1s orbital yielding the 2S1/2ground state of the neutral atom.
For I!, two spin–orbit states, significantly different inenergy, are accessed upon removal of a p-electron using267 nm radiation. The higher-eKE peak corresponds tophotoelectrons produced in coincidence with the ground(2P3/2) electronic state of the iodine atom, while the lowerpeak corresponds to the excited state (2P1/2). The 0.94 eVspacing between the peaks reflects the magnitude of thespin–orbit interaction in the iodine atom. The relativeintegrated peak intensities are proportional to the transitionprobabilities, including the degeneracies of the 2P3/2 and2P1/2 states of iodine.19 The finite peak widths in the energydomain reflect experimental broadening factors, such as laserlinewidth, parent-ion velocity spread, etc.16
3.2 The photoelectron angular distribution
The photoelectron angular distribution (PAD) is obtained byintegrating the intensity of the Abel-inverted image over achosen spectral (i.e. radial) range. By inspection, the H! PAD(Fig. 2f) has a ‘‘parallel’’ character, peaking at y = 0 and p,while the I! PADs (Fig. 3d) for both detachment channelspeak at y = p/2, corresponding to ‘‘perpendicular’’ photo-detachment transitions.The PADs in one-photon photodetachment with linearly
polarized light are generally described by the function:20
I(y) = (s/4p) [1 + bP2(cosy)], (2)
where the proportionality constant s corresponds to the totalphotodetachment cross-section and P2(cosy) " 1/2(3cos2y ! 1)is the second-order Legendre polynomial. The angulardependence is completely defined by b, the anisotropy para-meter, which can be determined by fitting eqn (2) to theexperimental PAD. The b values range from !1 to +2, withnegative and positive values corresponding to perpendicularand parallel polarization, respectively. The b value observed inH! photodetachment is very close to the +2 limit of a purelyparallel transition. For I! photodetachment, however, therings in Fig. 3(a) correspond to b E !0.6.
4. Interpreting photoelectron images
4.1 Relating photoelectron angular distributions to parentorbitals: the H! example
In a pulsed laser experiment the electron is described by asuperposition of waves spanning a finite energy range. In thelaboratory frame, this wavepacket is described as:
C(r,y,f,t) =Ra(k)Ck (r,y,f,t)dk (3)
where a(k) is a shape function related to the laser lineshape.Assuming a central, i.e. spherically symmetric, potentialCk(r,y,f,t) can be separated into radial and angular components,
Ck(r,y,f,t) = Rk(r,t)Y(y,f) (4)
The time dependence is carried in the radial function and thecorresponding electron probability density can be likened to apattern on a spherical balloon. As the ‘‘balloon’’ is inflated,the pattern expands, but its angular dependence remainsunchanged.The connection between a photoelectron image and the
parent orbital can be demonstrated using three different(yet closely inter-related) approaches based on: (1) angularmomentum conservation; (2) transition dipole momentoperations; (3) general symmetry considerations. In thefollowing, we adopt the one-electron picture of photo-detachment and discuss the uniquely simple case of H! toillustrate the key ideas in photoelectron image interpretation.
4.1.1 Conservation of angular momentum. The selectionrule for one-photon, one-electron atomic transitions isDc = #1, where c is the orbital angular momentum quantumnumber. In detachment from the hydrogen 1s orbital (ci = 0),the free-electron wave is characterized by a single c value of 1,corresponding to a p wave.
Fig. 3 Photoelectron imaging of I! at 267 nm. (a) The photoelectron
image. (b) Inverse Abel transformation. The two distinct rings corres-
pond to the different peaks in the photoelectron spectrum (c). The
photoelectron angular distribution (d) corresponding to the 2P3/2
transition peaks at p/2, in contrast to the H! detachment in Fig. 2
and contrary to classical expectations.
2172 | Chem. Soc. Rev., 2009, 38, 2169–2177 This journal is $c The Royal Society of Chemistry 2009
Publ
ished
on
12 M
ay 2
009.
Dow
nloa
ded
by U
nive
rsity
of S
outh
ern
Calif
orni
a on
13/
05/2
014
22:3
2:37
.
View Article Online
x =
x =
x =
d<latexit sha1_base64="7LzwEvygEmN0lm7keC+IFGCLzXs=">AAAB7XicbVBNSwMxEJ2tX7V+VT16CRbBU9mtBT0WvXisYD+gXUs2m21js8mSZIWy9D948aCIV/+PN/+NabsHbX0w8Hhvhpl5QcKZNq777RTW1jc2t4rbpZ3dvf2D8uFRW8tUEdoikkvVDbCmnAnaMsxw2k0UxXHAaScY38z8zhNVmklxbyYJ9WM8FCxiBBsrtfvJiD2Eg3LFrbpzoFXi5aQCOZqD8lc/lCSNqTCEY617npsYP8PKMMLptNRPNU0wGeMh7VkqcEy1n82vnaIzq4QoksqWMGiu/p7IcKz1JA5sZ4zNSC97M/E/r5ea6MrPmEhSQwVZLIpSjoxEs9dRyBQlhk8swUQxeysiI6wwMTagkg3BW355lbRrVe+iWrurVxrXeRxFOIFTOAcPLqEBt9CEFhB4hGd4hTdHOi/Ou/OxaC04+cwx/IHz+QOIW48Z</latexit>
µ<latexit sha1_base64="YUIm7Tsc5b5P2pMGLKWRa4Nw0cU=">AAAB6nicbVBNSwMxEJ3Ur1q/qh69BIvgqexWQY9FLx4r2g9ol5JNs21okl2SrFCW/gQvHhTx6i/y5r8xbfegrQ8GHu/NMDMvTAQ31vO+UWFtfWNzq7hd2tnd2z8oHx61TJxqypo0FrHuhMQwwRVrWm4F6ySaERkK1g7HtzO//cS04bF6tJOEBZIMFY84JdZJDz2Z9ssVr+rNgVeJn5MK5Gj0y1+9QUxTyZSlghjT9b3EBhnRllPBpqVealhC6JgMWddRRSQzQTY/dYrPnDLAUaxdKYvn6u+JjEhjJjJ0nZLYkVn2ZuJ/Xje10XWQcZWklim6WBSlAtsYz/7GA64ZtWLiCKGau1sxHRFNqHXplFwI/vLLq6RVq/oX1dr9ZaV+k8dRhBM4hXPw4QrqcAcNaAKFITzDK7whgV7QO/pYtBZQPnMMf4A+fwBerI3a</latexit>
el<latexit sha1_base64="PnLKfJSMMef3ecrAM3g5TVHWYXU=">AAAB8HicbVBNS8NAEJ34WetX1aOXxSJ4KkkV9Fj04rGC/ZA2ls120i7dbMLuRiihv8KLB0W8+nO8+W/ctjlo64OBx3szzMwLEsG1cd1vZ2V1bX1js7BV3N7Z3dsvHRw2dZwqhg0Wi1i1A6pRcIkNw43AdqKQRoHAVjC6mfqtJ1Sax/LejBP0IzqQPOSMGis9dOuaP2YoJr1S2a24M5Bl4uWkDDnqvdJXtx+zNEJpmKBadzw3MX5GleFM4KTYTTUmlI3oADuWShqh9rPZwRNyapU+CWNlSxoyU39PZDTSehwFtjOiZqgXvan4n9dJTXjlZ1wmqUHJ5ovCVBATk+n3pM8VMiPGllCmuL2VsCFVlBmbUdGG4C2+vEya1Yp3XqneXZRr13kcBTiGEzgDDy6hBrdQhwYwiOAZXuHNUc6L8+58zFtXnHzmCP7A+fwB/IOQhw==</latexit>x =
FIG. 3: Dipole selection rules in photoionization/photodetachment. Ionization/detachment from
an s-orbital yields a pure p-wave, giving rise to β=2, as in photodetachment from H−. Ion-
ization/detachment from a p-orbital yields interfering s- and d-waves, giving rise to a distinctly
different PAD. Images on the right show PADs for photodetachment from H− and I− (reproduced
from Ref. 27 with permission from the Royal Society of Chemistry).
Thus, PADs (or β’s) encode the information about the shapes of the Dyson orbitals from
which the photoelectrons originate27–29. PADs can be used to quantify the extent of orbital
deformation in the molecular environment due to hybridization and polarization. For ex-
ample, PADs of gas-phase water molecules reveal that the molecular Dyson orbitals can be
described as slightly deformed 2p(O) orbitals, giving rise to β ≈1.5 at high photoelectron
energies (above 100 eV). PADs are also used to interrogate the extent to which a solute’s
wavefunction is perturbed by the solvent29. A recent study30 used angular-resolved pho-
toelectron spectroscopy and the microjets technique to understand the electronic structure
of bulk water, i.e., to assess whether water molecules in bulk retain their electronic identi-
ties or give rise to qualitatively different delocalized band-like states. As shown in Fig. 4,
delocalized states are expected to give rise to an isotropic distribution of photoelectrons,
in contrast to β ≈1.5 of localized states of individual water molecules. So which picture
applies to bulk water? It turns out, the answer depends on the energy of the ionizing ra-
![Page 9: From orbitals to observables and backiopenshell.usc.edu/pubs/pdf/jcp_perspective.pdfQuantum mechanics explains chemical and physical phenomena by using abstract math-ematical constructs,](https://reader034.vdocuments.us/reader034/viewer/2022042406/5f1fa754210ddc366d060e82/html5/thumbnails/9.jpg)
9
diation: in the low-energy regime, one sees delocalized states, whereas in the high-energy
regime one observes the states of the individual waters. The two regimes and the transition
between them can be understood in terms of interference between photoelectrons coming
off individual molecules. When the de Broglie wavelength of photoelectrons is shorter than
the distance between the individual molecules (20 eV, corresponding to 3.5 A, first peak
in g(r) of the liquid water), then there is no interference, and one observes an incoherent
superposition of photoelectrons from the individual centers. Such interplay between the de
Broglie wavelength of photoelectrons and the distance between the emitting centers has been
illustrated by Sanov’s molecular interferometer experiment, showing the transition from the
delocalized and entangled molecular state of I−2 to the incoherent atomic states in the course
of dissociation31. Surprisingly, the microjet measurements of neat water30 show persistent
anisotropy of the photoelectrons, with only moderate reduction of β relative to the gas
phase, due to transport scattering. In stark contrast to the energetic consequences of hy-
drogen bonding and electrostatic interactions, manifested by large (1-2 eV) solvent-induced
shifts and inhomogeneous broadening of the valence bands, these interactions do not result
in reduction of anisotropy, meaning that the valence states of water in bulk can be thought
of as energy-shifted valence states of the isolated water molecules.
The utility of Dyson orbitals goes far beyond the interpretation of photoelectron or elec-
tron momentum (e, 2e) spectra. Because they show the difference between the N and N -1
electron states, the shape of Dyson orbitals can be related to the differences in electron den-
sity and, consequently, bonding patters. In the formaldehyde example (Fig. 2), the Dyson
orbital corresponding to the lowest ionization energy can be described as a lone pair on oxy-
gen (with smaller contributions from σOH). Thus, the ionization of this mostly non-bonding
orbital leads to relatively small changes in the geometry and short vibrational progression
(a small feature at the onset is due to the respective Franck–Condon factors).
In open-shell species (e.g., radicals), Dyson orbitals can be used to describe states of
the unpaired electrons. Fig. 5 shows Dyson orbitals of the four lowest electronic states
in SrOH. The orbitals have clear atomic-like characters, revealing that these species can
be described as a cationic core with the unpaired electron residing on the Sr. Because of
this feature, the electronic transitions in SrOH are atomic-like and the change in the state
of the unpaired electron does not perturb the core. Consequently, the geometry does not
change upon excitation, leading to diagonal Franck-Condon factors. This type of electronic
![Page 10: From orbitals to observables and backiopenshell.usc.edu/pubs/pdf/jcp_perspective.pdfQuantum mechanics explains chemical and physical phenomena by using abstract math-ematical constructs,](https://reader034.vdocuments.us/reader034/viewer/2022042406/5f1fa754210ddc366d060e82/html5/thumbnails/10.jpg)
10
-1.0
0.0
1.0
2.0
0 100 200 300 400 500 600 700
1b1,1<latexit sha1_base64="WY/7JU4JW55DZWOGtPyluyr0PT4=">AAAB9HicbVBNS8NAEJ3Ur1q/qh69LBbBg5SsCHosevFYwX5AG8Jmu2mXbjZxd1Moob/DiwdFvPpjvPlv3LQ5aOuDgcd7M8zMCxLBtXHdb6e0tr6xuVXeruzs7u0fVA+P2jpOFWUtGotYdQOimeCStQw3gnUTxUgUCNYJxne535kwpXksH800YV5EhpKHnBJjJa+fjLif4cDHF3jmV2tu3Z0DrRJckBoUaPrVr/4gpmnEpKGCaN3DbmK8jCjDqWCzSj/VLCF0TIasZ6kkEdNeNj96hs6sMkBhrGxJg+bq74mMRFpPo8B2RsSM9LKXi/95vdSEN17GZZIaJuliUZgKZGKUJ4AGXDFqxNQSQhW3tyI6IopQY3Oq2BDw8surpH1Zx24dP1zVGrdFHGU4gVM4BwzX0IB7aEILKDzBM7zCmzNxXpx352PRWnKKmWP4A+fzB8izkXA=</latexit><latexit sha1_base64="WY/7JU4JW55DZWOGtPyluyr0PT4=">AAAB9HicbVBNS8NAEJ3Ur1q/qh69LBbBg5SsCHosevFYwX5AG8Jmu2mXbjZxd1Moob/DiwdFvPpjvPlv3LQ5aOuDgcd7M8zMCxLBtXHdb6e0tr6xuVXeruzs7u0fVA+P2jpOFWUtGotYdQOimeCStQw3gnUTxUgUCNYJxne535kwpXksH800YV5EhpKHnBJjJa+fjLif4cDHF3jmV2tu3Z0DrRJckBoUaPrVr/4gpmnEpKGCaN3DbmK8jCjDqWCzSj/VLCF0TIasZ6kkEdNeNj96hs6sMkBhrGxJg+bq74mMRFpPo8B2RsSM9LKXi/95vdSEN17GZZIaJuliUZgKZGKUJ4AGXDFqxNQSQhW3tyI6IopQY3Oq2BDw8surpH1Zx24dP1zVGrdFHGU4gVM4BwzX0IB7aEILKDzBM7zCmzNxXpx352PRWnKKmWP4A+fzB8izkXA=</latexit><latexit sha1_base64="WY/7JU4JW55DZWOGtPyluyr0PT4=">AAAB9HicbVBNS8NAEJ3Ur1q/qh69LBbBg5SsCHosevFYwX5AG8Jmu2mXbjZxd1Moob/DiwdFvPpjvPlv3LQ5aOuDgcd7M8zMCxLBtXHdb6e0tr6xuVXeruzs7u0fVA+P2jpOFWUtGotYdQOimeCStQw3gnUTxUgUCNYJxne535kwpXksH800YV5EhpKHnBJjJa+fjLif4cDHF3jmV2tu3Z0DrRJckBoUaPrVr/4gpmnEpKGCaN3DbmK8jCjDqWCzSj/VLCF0TIasZ6kkEdNeNj96hs6sMkBhrGxJg+bq74mMRFpPo8B2RsSM9LKXi/95vdSEN17GZZIaJuliUZgKZGKUJ4AGXDFqxNQSQhW3tyI6IopQY3Oq2BDw8surpH1Zx24dP1zVGrdFHGU4gVM4BwzX0IB7aEILKDzBM7zCmzNxXpx352PRWnKKmWP4A+fzB8izkXA=</latexit><latexit sha1_base64="WY/7JU4JW55DZWOGtPyluyr0PT4=">AAAB9HicbVBNS8NAEJ3Ur1q/qh69LBbBg5SsCHosevFYwX5AG8Jmu2mXbjZxd1Moob/DiwdFvPpjvPlv3LQ5aOuDgcd7M8zMCxLBtXHdb6e0tr6xuVXeruzs7u0fVA+P2jpOFWUtGotYdQOimeCStQw3gnUTxUgUCNYJxne535kwpXksH800YV5EhpKHnBJjJa+fjLif4cDHF3jmV2tu3Z0DrRJckBoUaPrVr/4gpmnEpKGCaN3DbmK8jCjDqWCzSj/VLCF0TIasZ6kkEdNeNj96hs6sMkBhrGxJg+bq74mMRFpPo8B2RsSM9LKXi/95vdSEN17GZZIaJuliUZgKZGKUJ4AGXDFqxNQSQhW3tyI6IopQY3Oq2BDw8surpH1Zx24dP1zVGrdFHGU4gVM4BwzX0IB7aEILKDzBM7zCmzNxXpx352PRWnKKmWP4A+fzB8izkXA=</latexit>
1b1,2<latexit sha1_base64="PCNRlLaOamNzyyGd87NQEe3zig0=">AAAB9HicbVBNSwMxEJ2tX7V+VT16CRbBg5RNEfRY9OKxgv2AdlmyabYNzWa3SbZQlv4OLx4U8eqP8ea/MW33oK0PBh7vzTAzL0gE18Z1v53CxubW9k5xt7S3f3B4VD4+aek4VZQ1aSxi1QmIZoJL1jTcCNZJFCNRIFg7GN3P/faEKc1j+WSmCfMiMpA85JQYK3m9ZMj9DAc+vqrN/HLFrboLoHWCc1KBHA2//NXrxzSNmDRUEK272E2MlxFlOBVsVuqlmiWEjsiAdS2VJGLayxZHz9CFVfoojJUtadBC/T2RkUjraRTYzoiYoV715uJ/Xjc14a2XcZmkhkm6XBSmApkYzRNAfa4YNWJqCaGK21sRHRJFqLE5lWwIePXlddKqVbFbxY/XlfpdHkcRzuAcLgHDDdThARrQBApjeIZXeHMmzovz7nwsWwtOPnMKf+B8/gDKOJFx</latexit><latexit sha1_base64="PCNRlLaOamNzyyGd87NQEe3zig0=">AAAB9HicbVBNSwMxEJ2tX7V+VT16CRbBg5RNEfRY9OKxgv2AdlmyabYNzWa3SbZQlv4OLx4U8eqP8ea/MW33oK0PBh7vzTAzL0gE18Z1v53CxubW9k5xt7S3f3B4VD4+aek4VZQ1aSxi1QmIZoJL1jTcCNZJFCNRIFg7GN3P/faEKc1j+WSmCfMiMpA85JQYK3m9ZMj9DAc+vqrN/HLFrboLoHWCc1KBHA2//NXrxzSNmDRUEK272E2MlxFlOBVsVuqlmiWEjsiAdS2VJGLayxZHz9CFVfoojJUtadBC/T2RkUjraRTYzoiYoV715uJ/Xjc14a2XcZmkhkm6XBSmApkYzRNAfa4YNWJqCaGK21sRHRJFqLE5lWwIePXlddKqVbFbxY/XlfpdHkcRzuAcLgHDDdThARrQBApjeIZXeHMmzovz7nwsWwtOPnMKf+B8/gDKOJFx</latexit><latexit sha1_base64="PCNRlLaOamNzyyGd87NQEe3zig0=">AAAB9HicbVBNSwMxEJ2tX7V+VT16CRbBg5RNEfRY9OKxgv2AdlmyabYNzWa3SbZQlv4OLx4U8eqP8ea/MW33oK0PBh7vzTAzL0gE18Z1v53CxubW9k5xt7S3f3B4VD4+aek4VZQ1aSxi1QmIZoJL1jTcCNZJFCNRIFg7GN3P/faEKc1j+WSmCfMiMpA85JQYK3m9ZMj9DAc+vqrN/HLFrboLoHWCc1KBHA2//NXrxzSNmDRUEK272E2MlxFlOBVsVuqlmiWEjsiAdS2VJGLayxZHz9CFVfoojJUtadBC/T2RkUjraRTYzoiYoV715uJ/Xjc14a2XcZmkhkm6XBSmApkYzRNAfa4YNWJqCaGK21sRHRJFqLE5lWwIePXlddKqVbFbxY/XlfpdHkcRzuAcLgHDDdThARrQBApjeIZXeHMmzovz7nwsWwtOPnMKf+B8/gDKOJFx</latexit><latexit sha1_base64="PCNRlLaOamNzyyGd87NQEe3zig0=">AAAB9HicbVBNSwMxEJ2tX7V+VT16CRbBg5RNEfRY9OKxgv2AdlmyabYNzWa3SbZQlv4OLx4U8eqP8ea/MW33oK0PBh7vzTAzL0gE18Z1v53CxubW9k5xt7S3f3B4VD4+aek4VZQ1aSxi1QmIZoJL1jTcCNZJFCNRIFg7GN3P/faEKc1j+WSmCfMiMpA85JQYK3m9ZMj9DAc+vqrN/HLFrboLoHWCc1KBHA2//NXrxzSNmDRUEK272E2MlxFlOBVsVuqlmiWEjsiAdS2VJGLayxZHz9CFVfoojJUtadBC/T2RkUjraRTYzoiYoV715uJ/Xjc14a2XcZmkhkm6XBSmApkYzRNAfa4YNWJqCaGK21sRHRJFqLE5lWwIePXlddKqVbFbxY/XlfpdHkcRzuAcLgHDDdThARrQBApjeIZXeHMmzovz7nwsWwtOPnMKf+B8/gDKOJFx</latexit>
1p2
1b1,1 + 1b1,2
<latexit sha1_base64="mkO2MK53bSRVE3CNObzyWRy33e8=">AAACIHicbVDLSgMxFM34rPU16tJNsAgVpUyKUJdFNy4r2Ad0hiGTZtrQzMMkI5RhPsWNv+LGhSK6068xnc6ith64cHLOveTe48WcSWVZ38bK6tr6xmZpq7y9s7u3bx4cdmSUCELbJOKR6HlYUs5C2lZMcdqLBcWBx2nXG99M/e4jFZJF4b2axNQJ8DBkPiNYack1G7YvMElRltryQai0nmW2x4ZVOx4xN0Weiy5Qdj73quf+mWtWrJqVAy4TVJAKKNByzS97EJEkoKEiHEvZR1asnBQLxQinWdlOJI0xGeMh7Wsa4oBKJ80PzOCpVgbQj4SuUMFcnZ9IcSDlJPB0Z4DVSC56U/E/r58o/8pJWRgnioZk9pGfcKgiOE0LDpigRPGJJpgIpneFZIR1YkpnWtYhoMWTl0mnXkNWDd1dVprXRRwlcAxOQBUg0ABNcAtaoA0IeAIv4A28G8/Gq/FhfM5aV4xi5gj8gfHzC+b5ors=</latexit><latexit sha1_base64="mkO2MK53bSRVE3CNObzyWRy33e8=">AAACIHicbVDLSgMxFM34rPU16tJNsAgVpUyKUJdFNy4r2Ad0hiGTZtrQzMMkI5RhPsWNv+LGhSK6068xnc6ith64cHLOveTe48WcSWVZ38bK6tr6xmZpq7y9s7u3bx4cdmSUCELbJOKR6HlYUs5C2lZMcdqLBcWBx2nXG99M/e4jFZJF4b2axNQJ8DBkPiNYack1G7YvMElRltryQai0nmW2x4ZVOx4xN0Weiy5Qdj73quf+mWtWrJqVAy4TVJAKKNByzS97EJEkoKEiHEvZR1asnBQLxQinWdlOJI0xGeMh7Wsa4oBKJ80PzOCpVgbQj4SuUMFcnZ9IcSDlJPB0Z4DVSC56U/E/r58o/8pJWRgnioZk9pGfcKgiOE0LDpigRPGJJpgIpneFZIR1YkpnWtYhoMWTl0mnXkNWDd1dVprXRRwlcAxOQBUg0ABNcAtaoA0IeAIv4A28G8/Gq/FhfM5aV4xi5gj8gfHzC+b5ors=</latexit><latexit sha1_base64="mkO2MK53bSRVE3CNObzyWRy33e8=">AAACIHicbVDLSgMxFM34rPU16tJNsAgVpUyKUJdFNy4r2Ad0hiGTZtrQzMMkI5RhPsWNv+LGhSK6068xnc6ith64cHLOveTe48WcSWVZ38bK6tr6xmZpq7y9s7u3bx4cdmSUCELbJOKR6HlYUs5C2lZMcdqLBcWBx2nXG99M/e4jFZJF4b2axNQJ8DBkPiNYack1G7YvMElRltryQai0nmW2x4ZVOx4xN0Weiy5Qdj73quf+mWtWrJqVAy4TVJAKKNByzS97EJEkoKEiHEvZR1asnBQLxQinWdlOJI0xGeMh7Wsa4oBKJ80PzOCpVgbQj4SuUMFcnZ9IcSDlJPB0Z4DVSC56U/E/r58o/8pJWRgnioZk9pGfcKgiOE0LDpigRPGJJpgIpneFZIR1YkpnWtYhoMWTl0mnXkNWDd1dVprXRRwlcAxOQBUg0ABNcAtaoA0IeAIv4A28G8/Gq/FhfM5aV4xi5gj8gfHzC+b5ors=</latexit><latexit sha1_base64="mkO2MK53bSRVE3CNObzyWRy33e8=">AAACIHicbVDLSgMxFM34rPU16tJNsAgVpUyKUJdFNy4r2Ad0hiGTZtrQzMMkI5RhPsWNv+LGhSK6068xnc6ith64cHLOveTe48WcSWVZ38bK6tr6xmZpq7y9s7u3bx4cdmSUCELbJOKR6HlYUs5C2lZMcdqLBcWBx2nXG99M/e4jFZJF4b2axNQJ8DBkPiNYack1G7YvMElRltryQai0nmW2x4ZVOx4xN0Weiy5Qdj73quf+mWtWrJqVAy4TVJAKKNByzS97EJEkoKEiHEvZR1asnBQLxQinWdlOJI0xGeMh7Wsa4oBKJ80PzOCpVgbQj4SuUMFcnZ9IcSDlJPB0Z4DVSC56U/E/r58o/8pJWRgnioZk9pGfcKgiOE0LDpigRPGJJpgIpneFZIR1YkpnWtYhoMWTl0mnXkNWDd1dVprXRRwlcAxOQBUg0ABNcAtaoA0IeAIv4A28G8/Gq/FhfM5aV4xi5gj8gfHzC+b5ors=</latexit>
1p2
1b1,1 1b1,2
<latexit sha1_base64="3RhheM5pedeXCMzSKOjrv8214YQ=">AAACIHicbVDLSgMxFM34rPU16tJNsAgVtEyKUJdFNy4r2Ad0hiGTZtrQzMMkI5RhPsWNv+LGhSK6068xnc6ith64cHLOveTe48WcSWVZ38bK6tr6xmZpq7y9s7u3bx4cdmSUCELbJOKR6HlYUs5C2lZMcdqLBcWBx2nXG99M/e4jFZJF4b2axNQJ8DBkPiNYack1G7YvMElRltryQai0nmW2x4ZVOx4xN0Wei85RdjH3quf+mWtWrJqVAy4TVJAKKNByzS97EJEkoKEiHEvZR1asnBQLxQinWdlOJI0xGeMh7Wsa4oBKJ80PzOCpVgbQj4SuUMFcnZ9IcSDlJPB0Z4DVSC56U/E/r58o/8pJWRgnioZk9pGfcKgiOE0LDpigRPGJJpgIpneFZIR1YkpnWtYhoMWTl0mnXkNWDd1dVprXRRwlcAxOQBUg0ABNcAtaoA0IeAIv4A28G8/Gq/FhfM5aV4xi5gj8gfHzC+olor0=</latexit><latexit sha1_base64="3RhheM5pedeXCMzSKOjrv8214YQ=">AAACIHicbVDLSgMxFM34rPU16tJNsAgVtEyKUJdFNy4r2Ad0hiGTZtrQzMMkI5RhPsWNv+LGhSK6068xnc6ith64cHLOveTe48WcSWVZ38bK6tr6xmZpq7y9s7u3bx4cdmSUCELbJOKR6HlYUs5C2lZMcdqLBcWBx2nXG99M/e4jFZJF4b2axNQJ8DBkPiNYack1G7YvMElRltryQai0nmW2x4ZVOx4xN0Wei85RdjH3quf+mWtWrJqVAy4TVJAKKNByzS97EJEkoKEiHEvZR1asnBQLxQinWdlOJI0xGeMh7Wsa4oBKJ80PzOCpVgbQj4SuUMFcnZ9IcSDlJPB0Z4DVSC56U/E/r58o/8pJWRgnioZk9pGfcKgiOE0LDpigRPGJJpgIpneFZIR1YkpnWtYhoMWTl0mnXkNWDd1dVprXRRwlcAxOQBUg0ABNcAtaoA0IeAIv4A28G8/Gq/FhfM5aV4xi5gj8gfHzC+olor0=</latexit><latexit sha1_base64="3RhheM5pedeXCMzSKOjrv8214YQ=">AAACIHicbVDLSgMxFM34rPU16tJNsAgVtEyKUJdFNy4r2Ad0hiGTZtrQzMMkI5RhPsWNv+LGhSK6068xnc6ith64cHLOveTe48WcSWVZ38bK6tr6xmZpq7y9s7u3bx4cdmSUCELbJOKR6HlYUs5C2lZMcdqLBcWBx2nXG99M/e4jFZJF4b2axNQJ8DBkPiNYack1G7YvMElRltryQai0nmW2x4ZVOx4xN0Wei85RdjH3quf+mWtWrJqVAy4TVJAKKNByzS97EJEkoKEiHEvZR1asnBQLxQinWdlOJI0xGeMh7Wsa4oBKJ80PzOCpVgbQj4SuUMFcnZ9IcSDlJPB0Z4DVSC56U/E/r58o/8pJWRgnioZk9pGfcKgiOE0LDpigRPGJJpgIpneFZIR1YkpnWtYhoMWTl0mnXkNWDd1dVprXRRwlcAxOQBUg0ABNcAtaoA0IeAIv4A28G8/Gq/FhfM5aV4xi5gj8gfHzC+olor0=</latexit><latexit sha1_base64="3RhheM5pedeXCMzSKOjrv8214YQ=">AAACIHicbVDLSgMxFM34rPU16tJNsAgVtEyKUJdFNy4r2Ad0hiGTZtrQzMMkI5RhPsWNv+LGhSK6068xnc6ith64cHLOveTe48WcSWVZ38bK6tr6xmZpq7y9s7u3bx4cdmSUCELbJOKR6HlYUs5C2lZMcdqLBcWBx2nXG99M/e4jFZJF4b2axNQJ8DBkPiNYack1G7YvMElRltryQai0nmW2x4ZVOx4xN0Wei85RdjH3quf+mWtWrJqVAy4TVJAKKNByzS97EJEkoKEiHEvZR1asnBQLxQinWdlOJI0xGeMh7Wsa4oBKJ80PzOCpVgbQj4SuUMFcnZ9IcSDlJPB0Z4DVSC56U/E/r58o/8pJWRgnioZk9pGfcKgiOE0LDpigRPGJJpgIpneFZIR1YkpnWtYhoMWTl0mnXkNWDd1dVprXRRwlcAxOQBUg0ABNcAtaoA0IeAIv4A28G8/Gq/FhfM5aV4xi5gj8gfHzC+olor0=</latexit>
+ + + +
β
β 1
2
1
2
Kinetic energy [eV]A
niso
tropy
par
amet
er
and/or
A
B
D
C
FIG. 4: How do molecular orbitals of water clusters and bulk water look like? The answer is
encoded in photoelectron dipole anisotropy parameter β. For gas-phase water, β ≈1.5 at high
energies, as expected for ionization from a p-type orbital. Panel A shows model calculations of
water dimer: delocalized states treated with single-center expansion of the photoelectron state
give rise to isotropic distribution (β ≈ 0), whereas the treatment in terms of localized states
with no coherences results in β ≈1.5, as in isolated molecules. The domains of validity of the
two frameworks depend on de Broglie’s wave-length of photoelectrons in relation to the distance
between the centers. In high-energy regime, there is no interference between photoelectrons coming
off the individual waters, such that the resulting PAD reports on the valence orbitals of individual
waters in bulk (shown in panel B). Panels C and D illustrate that delocalized Dyson orbitals of bulk
water yield strikingly different results when treated within the single-center model (which ignores
the coherences in energy domain) and the multi-center model (which ignores the coherences in the
spatial domain). In the low-energy regime, delocalized picture is correct (panel C), whereas in the
high-energy regime, localized picture is correct (panel D). Reprinted with permission from Ref. 30.
Copyright (2020) American Chemical Society.
structure enables multiple optical transitions, which was instrumental in extending laser-
cooling techniques to polyatomic molecules33. The design of laser-coolable molecules has
been guided by the analysis of orbitals involved in optical transitions32,34–38, highlighting
the utility of molecular orbital theory in the emerging field of quantum information science.
![Page 11: From orbitals to observables and backiopenshell.usc.edu/pubs/pdf/jcp_perspective.pdfQuantum mechanics explains chemical and physical phenomena by using abstract math-ematical constructs,](https://reader034.vdocuments.us/reader034/viewer/2022042406/5f1fa754210ddc366d060e82/html5/thumbnails/11.jpg)
11
FIG. 5: Dyson orbitals showing the shape of the unpaired electron in SrOH. Reproduced from Ref.
32 with permission from the Royal Society of Chemistry.
B. Orbitals for excited states
It is difficult to imagine a conversation about excited states and photoinduced processes
without invoking molecular orbitals39,40. Yet, here again we face conceptual challenges: in
many-electron systems, electrons interact with each other, and changing the state of one
electron affects the states of the remaining ones. Hence, a naive Koopmans’ picture of
electronic excitations as removing an electron from an occupied orbital and attaching it to a
virtual orbital is wrong quantitatively and often even qualitatively. For example, in a typical
molecule estimates of the lowest excited-state energy as orbital energy gaps are off by many
electron-volts, due to a large electron–hole interaction energy neglected within the pseudo-
independent electron picture and a poor description of virtual orbitals. Qualitatively, the
character of excited states also often differs from orbital-energy based predictions, especially
when realistic basis sets with diffuse orbitals are used.
Within the lowest level of theory capturing the physics of excited states, configuration
interaction singles (CIS), the ground state is given by a Hartree–Fock determinant (Φ0)
and the target wavefunction is described as a linear combination of singly excited Slater
determinants, Φai :
ΨCIS =∑
ia
caiΦai . (8)
(Here we follow the standard convention: indices i, j, k, . . . denote orbitals occupied in the
reference determinant Φ0 and a, b, c, . . . denote orbitals from the virtual subspace.) The
analysis of the leading amplitudes cai in this expansion enables the interpretation of elec-
![Page 12: From orbitals to observables and backiopenshell.usc.edu/pubs/pdf/jcp_perspective.pdfQuantum mechanics explains chemical and physical phenomena by using abstract math-ematical constructs,](https://reader034.vdocuments.us/reader034/viewer/2022042406/5f1fa754210ddc366d060e82/html5/thumbnails/12.jpg)
12
tronic transitions in terms of hole and particle orbitals. For example, a CIS/6-31G* calcu-
lation of ethylene shows that the lowest excited state can indeed be described as a π → π∗
transition (the weight of the respective Φai ≈0.94), as expected from Koopmans’ theorem
(HOMO→LUMO transition); this is illustrated in Fig. 6. Unfortunately, this type of anal-
ysis (based on the wavefunction amplitudes) is prone to several problems. First, it is not
unique. The CIS energy and excited-state properties are invariant with respect to any
unitary transformation of orbitals within the occupied or virtual orbital spaces, but the
wavefunction amplitudes are not. Thus, choosing different orbitals changes the apparent
character of the wavefunction. One consequence of this is that the weights of the leading
amplitudes are strongly basis-set dependent. Often, in large bases the wavefunction features
several amplitudes with similar weights, which may or may not reflect the true character
of the transition. Second, this analysis quickly becomes intractable when both states ac-
quire multi-configurational character. For example, already at the CIS level, the analysis
of the transitions between two excited states is much more challenging than between the
ground and excited states. In correlated methods, both the ground-state and excited-state
wavefunctions become multiconfigurational and include multiply excited determinants.
To find a way out of this predicament, we begin by considering a relevant experimental
observable. The wavefunction amplitudes are not observables, but the transition density
is41. The transition density is an object that connects the two states:
ρ(xe, xh) =∑
pq
γFIpq φp(xe)φq(xh), (9)
where xe and xh denote electron (particle) and hole coordinates and γFIpq is a one-particle
transition density matrix between the initial and final states:
γFIpq = 〈ΨF |p†q|ΨI〉. (10)
Here p† and q are creation and annihilation operators corresponding to the φp and φq molec-
ular orbitals. The transition density ρ(xe, xh) is related to observable properties because it
defines the probability of transition. For example, the transition dipole moment between
two many-body states is:
〈ΨF |µ|ΨI〉 =
∫ρ(xe, xh)|xe=xh=rrdr = Tr[γFIµ]. (11)
In deriving this expression, we again used electron indistinguishability, following the same
steps as above. Because ρ(xe, xh) is mapped into the observable, it can, in principle be
![Page 13: From orbitals to observables and backiopenshell.usc.edu/pubs/pdf/jcp_perspective.pdfQuantum mechanics explains chemical and physical phenomena by using abstract math-ematical constructs,](https://reader034.vdocuments.us/reader034/viewer/2022042406/5f1fa754210ddc366d060e82/html5/thumbnails/13.jpg)
13
FIG. 6: Lowest excited triplet state in ethylene. In a small basis set (6-31G*), the CIS wavefunction
is dominated by the HOMO→LUMO excitation (the weight of the leading amplitude is 0.94), and
can be easily identified as the π → π∗ transition (the canonical Hartree–Fock MOs are shown on
the left). When the basis is extended to 6-31(2+)G*, several Rydberg orbitals appear below the π∗
valence MO and the CIS wavefunction of the lowest triplet state has two comparable amplitudes
involving both Rydberg and valence like orbitals. What is the true character of this state? The
NTO analysis reveals that the character of the lowest triplet state in the 6-31(2+)G* basis is
the same as in the 6-31G* basis: 3ππ∗. Rydberg triplet state, corresponding to the π → Ry(s)
transition, appears several electron-volt above 3ππ∗. In the singlet manifold, the ordering of the
π → Ry(s) and π → π∗ states is reversed.
reconstructed from the experimental data. When plotted in spatial coordinates, ρ(xe, xh)
provides a visual map of the changes in the electron density. By virtue of Eq. (11), one
can interpret ρ(xe, xh) as an exciton’s wavefunction and use it to rationalize the computed
properties42. For example, the shape of ρ(xe, xh) for different excited states in a bacte-
riochlorophyll dimer shown in Fig. 7 explains the relative intensities and polarization of
the main spectral features and clearly distinguishes local, entangled, and charge-transfer
transitions43.
γFI , the one-particle transition density matrix (1PTDM), contains the coefficients of the
expansion of ρ over molecular orbitals. It is a much simpler object than two many-body
wavefunctions. One can think of it as of a map connecting the two states by moving one
electron at a time, thus extending the CIS-like picture to correlated states:
ΨF =∑
pq
γFIpq p†qΨI + higher excitations. (12)
![Page 14: From orbitals to observables and backiopenshell.usc.edu/pubs/pdf/jcp_perspective.pdfQuantum mechanics explains chemical and physical phenomena by using abstract math-ematical constructs,](https://reader034.vdocuments.us/reader034/viewer/2022042406/5f1fa754210ddc366d060e82/html5/thumbnails/14.jpg)
14
charge-transfer excitations. These states lie just above the Qxtransitions and form a band that overlaps with the Qx Frenkel-exciton band in higher aggregates. Even though the CT modesare essentially forbidden in linear absorption, they show up innonlinear optical studies of these aggregates (e.g., electroab-sorption).We have performed a similar analysis of the higher frequency
Tx1-Tx3 and By transitions (see the second column of Table2), but in the present paper we limit our discussion to the low-frequency Q, Bx, and CT states that participate in the light-harvesting function of LH2.Finally we compare several different Bchl-a dimers with
strong electronic couplings. The third column of Table 2 showsthe 2R1! B850 aggregate (intersubunit). Since the intersubunitand intrasubunit aggregates are similar, the excitation energies
and splittings of these dimers (second and third column) arevery close. In contrast, a much weaker interaction is found indimers 2RB850-1B800 (upper-lower ring) and 2B800-1B800(lower-lower ring) since the center-to-center separation of ∼25Å is larger compared to that of the neighboring molecules ofthe upper ring. Consequently the CT modes of these aggregateslie∼1 eV higher in energy (compare columns 2, 3 with columns4, 5 of Table 2). Additional Bchl dimers and their electroniccouplings will be discussed in the next section.
V. Frenkel-Exciton Hamiltonian for LH2
The CEO analysis of the dimer’s electronic modes presentedin the previous section shows that the interaction between
Figure 3. (A) Structure of the R! B850 dimer. (B) Calculated linear absorption spectrum of the R! B850 dimer. Contour plots the electronicmodes that dominate the optical absorption of the B850 dimer. The axis labels represent the individual atoms. Atoms of each monomer are labeledaccording to panel A of Figure 2. The panels indicate the electronic mode according to panel B. The color code is given in Figure 2. Modefrequencies are given in Table 2.
4524 J. Phys. Chem. B, Vol. 104, No. 18, 2000 Tretiak et al.
FIG. 7: Analysis of excited states using transition density matrices in a bacteriochlorophyll dimer
from the LH2 antenna complex of purple bacteria. (A) Structure of the dimer. (B) Calculated
linear (i.e., one-photon) absorption spectrum of the dimer. Contour plots of the transition densities
corresponding to the main spectral transitions (the first panel shows the ground-state density).
The axis labels represent the individual atoms. The plots clearly show the extent of exciton
delocalization. Charge-transfer transitions (CT1 and CT2) are revealed by the off-diagonal pattern.
Reprinted with permission from Ref. 43. Copyright (2000) American Chemical Society.
Although it is not sufficient for complete reconstruction of ΨF from ΨI , γ contains all
the information needed to compute the transition dipole moment and, hence, relates to an
![Page 15: From orbitals to observables and backiopenshell.usc.edu/pubs/pdf/jcp_perspective.pdfQuantum mechanics explains chemical and physical phenomena by using abstract math-ematical constructs,](https://reader034.vdocuments.us/reader034/viewer/2022042406/5f1fa754210ddc366d060e82/html5/thumbnails/15.jpg)
15
experimentally observable property. For the CIS transitions from the ground state, i.e., when
ΨF = ΨCIS and ΨI = Φ0, only the occupied-virtual block of γ is non-zero and γia = cai .
The difference between two electronic states can be also described by density difference44:
∆(r) =∑
pq
∆FIpq φp(r)φq(r), (13)
∆FIpq = 〈ΨF |p†q|ΨF 〉 − 〈ΨI |p†q|ΨI〉. (14)
This quantity provides an alternative measure of the change in electronic structure upon
the transition and is related to the change in the one-electron state properties (such as
permanent dipole moment); hence, it is also related to observables. At the CIS level (for
the transitions from the ground state), γ and ∆ are closely related and contain identical
information45; however, for correlated wavefunctions (or transitions between the CIS states)
this is no longer the case. For example, ∆ can be used to characterize the differences between
the states that are multiply excited with respect to each other, whereas γ only captures the
differences in terms of one-electron excitations.
While ρ(xe, xh) and ∆(r) are invariant with respect to different orbital choices—any set
of orthonormal orbitals can be used in Eqs. (11)-(14)—the values of the individual elements
of γpq and ∆pq depend on the specific orbital choice in the same way the wavefunction
amplitudes do. Rather than obscuring the intrinsic nature of electronic transitions, this
arbitrariness is the key to deriving a rigorous and clear orbital picture. One can find the
best set of orbitals, which represents ∆pq and γFI in the most compact way.
An important difference between the two objects is that the state density matrix is Her-
mitian, whereas the transition density matrix is not46. Hence, the most compact representa-
tion of ∆pq is achieved by simple diagonalization, yielding natural density-difference orbitals
(NDOs) and their occupations, which add up to zero. By interpreting negative occupations
as electron detachment from state ΨI and positive occupations as electron attachment to
state ΨF , one can compute promotion numbers, which quantify how many electrons need
to be simultaneously excited to describe the transition. NDOs and their occupation num-
bers yield attachment and detachment densities, which compactly represent the difference
between the two states44,47.
In the case of the non-Hermitian 1PTDM, γFI , the most compact representation is ac-
complished by the singular value decomposition
γ = VΣUT, (15)
![Page 16: From orbitals to observables and backiopenshell.usc.edu/pubs/pdf/jcp_perspective.pdfQuantum mechanics explains chemical and physical phenomena by using abstract math-ematical constructs,](https://reader034.vdocuments.us/reader034/viewer/2022042406/5f1fa754210ddc366d060e82/html5/thumbnails/16.jpg)
16
where Σ is the diagonal matrix of singular values, σK , and matrices V and U contain the
hole and particle (electron) NTOs according to
ψhK (r) =∑
q
UqKφq (r) (16)
and
ψeK (r) =∑
q
VqKφq (r) . (17)
(Note that hole and particle orbitals are not orthogonal to each other). In the NTO repre-
sentation, ρ has the most compact form:
ρ(xe, xh) =∑
K
σKψeK(xe)ψ
hK(xh). (18)
Usually, only a few σK are non-vanishing. The squares of the σKs can be interpreted as the
weights of the respective NTO pair when divided by the square of the Frobenius norm (Ω)
of γ:
Ω ≡ ||γ||2 =∑
pq
γ2pq =∑
k
σ2K . (19)
Ω is also useful for identifying transitions that have predominantly one-electron character
(Ω=1 for a pure single-electron transition, such as between the ground state and a CIS
excited state)48.
NTOs are an old and extremely useful concept44,45,47,49–57, which provide a way to describe
electronic excitations in terms of molecular orbitals for any type of wavefunction. NTOs are
also directly related to the observables:
〈ΨF |µ|ΨI〉 = Tr[γµ] =∑
K
σK〈ψeK |µ|ψhK〉, (20)
that is, the matrix element between the two many-body states can be expressed by a (short)
sum of matrix elements between the hole and particle NTOs. This allows one to use an
orbital picture to explain trends. For example, the orbitals shown in Fig. 6 make it immedi-
ately clear that in the singlet manifold both types of transitions carry noticeable oscillator
strength, but are polarized differently: π → π∗ has the transition dipole along the molecular
axis, whereas π → Ry(3s) has the transition dipole perpendicular to the molecular plane.
Fig. 8 illustrates the ability of NTOs to unambiguously determine the character of the
excited states. The CIS/aug-cc-pVDZ wavefunction of the lowest singlet state features 5
![Page 17: From orbitals to observables and backiopenshell.usc.edu/pubs/pdf/jcp_perspective.pdfQuantum mechanics explains chemical and physical phenomena by using abstract math-ematical constructs,](https://reader034.vdocuments.us/reader034/viewer/2022042406/5f1fa754210ddc366d060e82/html5/thumbnails/17.jpg)
17
leading amplitudes (which collectively account for only 75% of the wavefunction), suggesting
a naive interpretation of this transition as having multi-configurational Rydberg-valence
character. However, the NTOs computed for this wavefunction reveal a pure (99.1 %)
valence lp(O) → π∗ transition. Thus, the mixed multi-configurational appearance in the
basis of canonical orbitals is just an artifact of the representation.
0.991
0.069
0.305 0.149 0.120 0.108
FIG. 8: Lowest excited singlet state in uracil, CIS/aug-cc-pVDZ. Top: weights of the leading
CIS amplitudes in the basis of canonical Hartree–Fock orbitals show multiple configurations with
similar weights (these configurations account for 75% of the wavefunction). Bottom: NTO analysis
of this wavefunction reveals simple lp(O) → π∗ valence transition (99.1 % of the wavefunction is
accounted for by this NTO pair).
The ability of NTOs to provide an essential picture of the transition becomes even more
important when using correlated wavefunctions. For example, in EOM-CCSD, both the
ground-state and excited-state wavefunctions include single and double excitations, making
amplitude-based analysis even more ambiguous. Yet, NTOs distill these complex wavefunc-
tions into a simple orbital picture. Fig. 9 shows NTOs for the bright XAg →1Bu transition
and dark XAg →4Ag transition in trans-stilbene computed using EOM-CCSD/daug-cc-
pVDZ wavefunctions. The analysis shows that 1Bu can be described as a simple π → π∗
excitation, whereas 4Ag has more complex character and requires 2 pairs of NTOs. Being
![Page 18: From orbitals to observables and backiopenshell.usc.edu/pubs/pdf/jcp_perspective.pdfQuantum mechanics explains chemical and physical phenomena by using abstract math-ematical constructs,](https://reader034.vdocuments.us/reader034/viewer/2022042406/5f1fa754210ddc366d060e82/html5/thumbnails/18.jpg)
18
4Ag
Trans stilbene, EOM-CCSD/daug-cc-pVDZ basis:
0.70
1Bu
0.54
0.21
FIG. 9: Trans-stilbene, EOM-CCSD/daug-cc-pVDZ. NTO analysis of the bright XAg →1Bu tran-
sition (top) and dark XAg →4Ag transition (bottom). NTO weights (σ2K) are shown for each NTO
pair.
independent of the correlation treatment and the basis, NTOs enable comparisons of wave-
functions computed by different methods55, including comparisons between wavefunction
treatments and TD-DFT.
The number of NTOs with significant weights reveals the inherent collectivity of the tran-
sition and can be used to quantify quantum entanglement of the exciton. Several collectivity
indices are used. For example, the participation ratio (PRNTO) is defined as:
PRNTO =(∑
K σ2K)2∑
K σ4K
=Ω2
∑K σ
4K
(21)
where Ω = ||γ||2. For a transition between a reference and a CIS target state, PRNTO equals
1 for a CIS target state with exactly one configuration, 2 for a CIS state represented by two
configurations with equal weights, and so on. Thus, pure excitations (such as ππ∗ or π →Ry(s) transitions in ethylene) are revealed by PRNTO ≈1, whereas states of mixed character
(e.g., Rydberg-valence) would have larger values. Plasmonic excitations are identified by
very large PRNTO.
PRNTO is closely related to the number of entangled states, an alternative metric58 also
![Page 19: From orbitals to observables and backiopenshell.usc.edu/pubs/pdf/jcp_perspective.pdfQuantum mechanics explains chemical and physical phenomena by using abstract math-ematical constructs,](https://reader034.vdocuments.us/reader034/viewer/2022042406/5f1fa754210ddc366d060e82/html5/thumbnails/19.jpg)
19
defined on the basis of the 1PTDMs:
ZHE = 2SHE = 1/∏
i
λλii , (22)
where λi = σ2i /Ω are renormalized squares of the singular values of 1PTDM and SHE is the
hole-electron entanglement entropy:
SHE = −∑
i
λi log2 λi. (23)
This definition gives the same answer in the idealized example above, but has an advantage
of being grounded in thermodynamics, which gives rise to the correct scaling with the system
size. For example, utilizing the additivity of entropy allows one to properly account for spin
entanglement in spin-adapted calculations (i.e., the total SHE is the sum of the αα and
ββ parts of the transition). Thus, for singlet–singlet transitions ZHE accounting for spin-
entanglement is the square of ZHE computed from the spinless transition density. A recent
study36 of novel molecular frameworks for quantum information science used these metrics
to investigate the entanglement between two optical cycling centers attached to the same
molecular scaffold. The calculations on model systems (Me-O-CC-OMe′, Me=Mg, Ca, Ba,
Sr) showed that the extent of entanglement and electronic communication can be controlled
by the diradical character.
NTOs obtained from 1PTDM can always be used to describe the difference between the
two states, however, their connection to spectroscopic observables requires further discussion.
As per Eq. (20), NTOs can be used to understand the trends in one-particle transition
properties. For example, from Fig. 9, one can easily see why the XAg →1Bu transition is
bright and the XAg →4Ag is dark in linear (UV-VIS) spectroscopy. However, it is not at
all obvious why the XAg →4Ag has a very large two-photon absorption (2PA) cross section,
simply because the NTOs shown in Fig. 9 do not directly relate to the 2PA cross section.
The probability of a two-photon transition between two states depends not only on the
initial and final state, but also on all other states of the system. For example, the 2PA
transition moments are given by the following sum-over-states (SOS) expressions59–61:
M f←ixy (ω1x, ω2y) = −
∑
n
(〈Ψf |µy|Ψn〉〈Ψn|µx|Ψi〉Ωni − ω1x
+〈Ψf |µx|Ψn〉〈Ψn|µy|Ψi〉
Ωni − ω2y
), (24)
M i←fxy (−ω1x,−ω2y) = . . .
![Page 20: From orbitals to observables and backiopenshell.usc.edu/pubs/pdf/jcp_perspective.pdfQuantum mechanics explains chemical and physical phenomena by using abstract math-ematical constructs,](https://reader034.vdocuments.us/reader034/viewer/2022042406/5f1fa754210ddc366d060e82/html5/thumbnails/20.jpg)
20
Here, Ωni = En−Ei and ω1 and ω2 are the frequencies of the two absorbed photons (polarized
along x- and y-directions) such that ω1 + ω2 = Ωfi.
Eq. (24) reflects important physics of the 2PA and other nonlinear optical processes—the
properties of the transition can no longer be understood by considering only the initial and
final states. Rather, one needs to consider all states of the system. Except for resonant
regimes (such as a 2PA transition in which one of the photons is resonant with a bright
excited state Ψn), the SOS expressions cannot be truncated to a few terms without significant
loss of accuracy, presenting an obstacle to deriving molecular orbital pictures of non-linear
phenomena, such as 2PA or RIXS (resonant inelastic X-ray scattering).
This obstacle can be overcome by exploiting response theory, the same approach used
to circumvent the calculation of all excited states entering the SOS expressions in practical
calculations of the 2PA and RIXS cross sections. By using response theory, Eqs. (24) can
be recast as:
M f←ixy (ω1x) = −
(〈Ψf |µy|Xω1x
i 〉+ 〈Xω1xf |µy|Ψi〉
), (25)
M i←fxy (−ω2y) = . . .
where the response states, X and X, are computed by solving auxiliary response
equations60–63. Within the formalism outlined in Refs. 61 and 63, the EOM-CC response
states are:
Xω1xk (H − Ek + ω1x) = 〈Ψk|µx|ΦI〉, (26)
(H − Ek − ω1x)Xω1xk = 〈ΦI |µx|Ψk〉, (27)
where ΦIs denote Slater determinants from the target-state manifold. The physical mean-
ing of response states becomes clear once we recognize that they describe the first-order
response64 of Ψk to the perturbation with frequency ωx:
|Xω1xk 〉 ≡ |Ψ
(1)k 〉 =
∑
n
|Ψn〉〈Ψn|µx|Ψk〉Ωnk − ω1x
, (28)
〈Xω1xk | ≡ 〈Ψ
(1)k | =
∑
n
〈Ψn|µx|Ψk〉Ωnk + ω1x
〈Ψn|. (29)
Hence, one can express (see Eq. (25)) 2PA cross sections as matrix elements of the dipole
operator between the initial state and the first-order perturbed final state plus the matrix
element between the perturbed initial state and the final state. The resulting approach
![Page 21: From orbitals to observables and backiopenshell.usc.edu/pubs/pdf/jcp_perspective.pdfQuantum mechanics explains chemical and physical phenomena by using abstract math-ematical constructs,](https://reader034.vdocuments.us/reader034/viewer/2022042406/5f1fa754210ddc366d060e82/html5/thumbnails/21.jpg)
21
is formally and numerically equivalent to the SOS expressions. The perturbed states X,
which assimilate the contributions from all real excited states of the system, provide con-
crete meaning to the so-called “virtual states”, which are commonly invoked in non-linear
spectroscopies.
Following Eq. (25), one can define and compute the respective 1PTDMs (called perturbed
1PTDMs) and extract the NTOs. Such analysis delivers simple molecular orbital pictures
of nonlinear processes. Fig. 10 shows NTOs for the bright 2PA transition in ethylene
(XAg →2Ag). As one can see from the regular (linear or one-photon) NTOs, the 2Ag state
has πRy(pz) character. The NTO analysis of the Mzz and Mxx 2PA moments shows that
the former moment is dominated by the pathway via the Ry(s) virtual state, whereas the
latter moment involves the π∗ virtual state.
Recently, following the same steps, we extended the NTO analysis of perturbed density
matrices to the RIXS process65. Fig. 11 shows the orbital picture of the RIXS transitions
in benzene derived from the NTOs computed from perturbed density matrices. As in the
case of 2PA, these NTOs are directly mapped onto the cross sections. Without making
simplifying assumptions, they reveal the character of the virtual state responsible for the
transitions. In the benzene example, the character of the virtual state changes from valence
π∗-like to Rydberg s-like depending on the frequency of the excitation pulse.
![Page 22: From orbitals to observables and backiopenshell.usc.edu/pubs/pdf/jcp_perspective.pdfQuantum mechanics explains chemical and physical phenomena by using abstract math-ematical constructs,](https://reader034.vdocuments.us/reader034/viewer/2022042406/5f1fa754210ddc366d060e82/html5/thumbnails/22.jpg)
22
One-photon NTOsTwo-photon NTOs (Mzz)
Two-photon NTOs (Mxx)
One-photon NTOsTwo-photon NTOs (Mzz)
Two-photon NTOs (Mxx)
FIG. 10: One-photon (top) and two-photon (bottom) NTOs for the XAg →2Ag transition in
ethylene; EOM-CCSD/daug-cc-pVDZ. The final excited state can be described as Ry(pz). 2PA
NTOs reveal virtual states responsible for 2PA: Mzz and Mxx moments can be described as π →
Ry(s)→ Ry(pz) and π → π∗ → Ry(pz) transitions, respectively.
![Page 23: From orbitals to observables and backiopenshell.usc.edu/pubs/pdf/jcp_perspective.pdfQuantum mechanics explains chemical and physical phenomena by using abstract math-ematical constructs,](https://reader034.vdocuments.us/reader034/viewer/2022042406/5f1fa754210ddc366d060e82/html5/thumbnails/23.jpg)
23
15
virtual molecular orbitals (not shown) are doubly degenerate LUMO and di↵use s and p
Rydberg orbitals.
The two dominant features in the XAS spectrum of benzene42–49 are peak A and peak B at
285.97 eV and 287.80 eV, respectively (theoretical values21 computed with fc-CVS-EOM-EE-
CCSD/uC-6-311(2+,+)G**). When the incoming photon’s energy is tuned to the peak-A
resonance, the dominant inelastic feature is the energy-loss peak at 10.67 eV, characterized
by equal contributions from the degenerate XAg ! 13B2g and XAg ! 12B3g transitions. In
contrast, when the incoming photon’s energy is tuned to the peak-B resonance, the dominant
inelastic feature is the energy-loss peak at 6.45 eV, characterized by equal contributions
from the degenerate XAg ! 1B2g and XAg ! 1B3g transitions. Below we show the NTO
analysis of only the XAg ! 13B2g and XAg ! 1B2g RIXS transitions with incoming photon
energies tuned at peak-A and peak-B resonances, respectively. The NTO analyses for the
two transitions are similar except for the di↵erences in symmetry labels of the orbitals.
The NTO analysis for the dark one-photon XAg ! 13B2g transition (given in the SI)
suggests that this valence transition is made up of two orbital transitions: b2u ! au and
b3u ! b1u. Similarly, the NTO analysis of XAS peak A (XAg ! c2B1u) transition in
Fig. 6 shows two dominant orbital transitions: cb1g ! au and cag ! b1u. Similarly, the
FIG. 6: NTO analysis of the XAg ! 13B2g RIXS transition in benzene. Comparison of important
NTO pairs computed separately for the x-ray absorption (XAg ! c2B1u) and x-ray emission
(c2B1u ! 13B2g) transitions with the NTO pairs computed from the RIXS 1PTDMs shows that
the three-states model is adequate for this RIXS transition. Both orbital channels contribute
significantly into this transition.
16
NTO analysis of the c2B1u ! 13B2g x-ray emission shows two dominant orbital transitions:
b2u ! cb1g and b3u ! cag. Based on these analyses, the three-states model for the XAg !13B2g RIXS transition identifies two important orbital channels: b2u ! cb1g ! au and
b3u ! cag ! b1u.
The g ! f and f ! g RIXS moment tensors are dominated by the imaginary zx
components. This is also reflected in the norms of the imaginary 1PTDMs given in the SI,
which are a few orders of magnitude larger than those of real 1PTDMs (Im = 1.00). The
NTO analyses of the z,Im and x,Im RIXS 1PTDMs of the XAg ! 13B2g shown in Fig.
6 and in the SI identifies two dominant near-resonance orbital channels: b2u ! cb1g ! au
and b3u ! cag ! b1u. In other words, cb1g and cag are the intermediate core orbitals that
facilitate the two-photon inelastic scattering, driving the electronic density from the b2u
orbital to the 2au orbital and from the b3u orbital to the 4b1u orbital, respectively. This
orbital analysis of the RIXS 1PTDMs is consistent with the approximate analysis from the
three-states model.
The one-photon XAg ! 1B2g transition is dark; its NTO analysis given in the SI reveals
its dominant HOMO-LUMO character (b2g ! ag) and a miniscule contribution from the
b1u ! b3u transition. The NTO analysis of XAS peak B transition (XAg ! c2B3u) shown
FIG. 7: NTO analysis of the 1Ag ! 1B2g RIXS transition in benzene. Comparison of impor-
tant NTO pairs computed separately for the x-ray absorption (1Ag ! c2B3u) and x-ray emission
(c2B3u ! 1B2g) transitions with the NTO pairs computed from the RIXS 1PTDMs shows that
the three-states model is adequate for this RIXS transition. Orbital channel 1 provides dominant
contributions.
FIG. 11: NTO analysis of perturbed densities reveals different channels giving rise to RIXS emission
in benzene. Top: Orbital picture of RIXS transition due to resonant excitation at 286 eV. Bottom:
Orbital picture of RIXS transition due to resonant excitation at 288 eV. Reprinted from Ref. 65,
with the permission of AIP Publishing.
![Page 24: From orbitals to observables and backiopenshell.usc.edu/pubs/pdf/jcp_perspective.pdfQuantum mechanics explains chemical and physical phenomena by using abstract math-ematical constructs,](https://reader034.vdocuments.us/reader034/viewer/2022042406/5f1fa754210ddc366d060e82/html5/thumbnails/24.jpg)
24
C. Orbital picture of spin-forbidden transitions and magnetic properties
Molecular orbital theory has been used to explain relativistic phenomena, ranging from
orbital contraction in heavy elements due to scalar relativistic effects to inter-system crossing
(ISC), intensity borrowing, and magnetic anisotropies facilitated by spin-orbit interactions.
The last effect plays an important role even in molecules composed of light atoms, as it cou-
ples otherwise non-interacting states (e.g., singlets and triplets), opening up new relaxation
pathways.
In photochemistry, the rates of ISC are often explained by El-Sayed’s rules66,67, which
state that a large SOC (spin-orbit coupling) can be expected for transitions involving a
flip of an orbital, such as in [(π)2(n)1(π∗)1] → [(π)2(n)2(π∗)0], whereas transitions that
do not involve orbital flip, such as in [(π)1(n)2(π∗)1] → [(π)2(n)2(π∗)0], would have much
smaller SOC and, consequently, slower ISC. El-Sayed’s rules are based on the analysis of
the one-electron part of the Breit–Pauli Hamiltonian, which is responsible for the SOC. As
illustrated in Fig. 12, the SOC term has the form of a local angular momentum operator,
whose matrix representation reveals the need for orbital torque—for example, in the basis
of three p-orbitals, the only non-zero matrix elements of Lz are between px and py.
Can we use the concept of NTOs to develop a quantitative version of El-Sayed’s rules
applicable to correlated many-body wavefunctions? The conceptual difficulty is that the
SOC is a tensorial property that depends on the matrix elements between all multiplet
components. For example, the SOC constant (SOCC), which enters the expressions for ISC
rates and spectral line splittings, involves the sum over all spin-projections within the two
coupled multiplets (e.g., the three Ms components of a triplet state, in the case of singlet-
triplet couplings). Thus, calculation of NTOs between one multiplet component (such as
the Ms=0 component, most often available from electronic structure calculations) is not
sufficient for quantitative analysis of the SOCC. This obstacle can be overcome by using
Wigner-Eckart’s theorem and introducing a new type of NTOs appropriate for the analysis
of spin-forbidden phenomena68. The definition of spin-orbit NTOs68 is based on the spinless
density matrices that are used for the calculation of the SOCs for the entire multiplet from
just one transition69,70. This approach allows one to treat the transitions between states with
arbitrary spin projections in a uniform way and to quantitatively describe the contributions
of specific orbital pairs to the overall SOC; hence, these NTOs are linked to experimentally
![Page 25: From orbitals to observables and backiopenshell.usc.edu/pubs/pdf/jcp_perspective.pdfQuantum mechanics explains chemical and physical phenomena by using abstract math-ematical constructs,](https://reader034.vdocuments.us/reader034/viewer/2022042406/5f1fa754210ddc366d060e82/html5/thumbnails/25.jpg)
25
observable quantities (e.g., ISC and oscillator strengths). Fig. 12 shows NTOs computed for
the SOC within the quintet manifold in an iron-contained single-molecule magnet described
by the EOM-EA-CC ansatz68,70. A large spin-reversal barrier71 of this single-molecule mag-
net can be explained by the shapes of these NTOs. In the spirit of El-Sayed’s rule, these
two orbitals, which can be described as dzy and dz2 , give rise to non-zero matrix elements of
the angular momentum operator.
Analysis: Extended NTOs
Example: tris(pyrrolylmethyl)amine Fe(II) complex,quintet-quintet transition, EOM-EA-MP2
ω=0.87
cm1 NTO integrals ! Reduced matrix elements (exact)
hS||HSOL ||Si 31.79 + 224.48i 32.07 + 228.23i
hS||HSOL0
||Si 0.04i 0.07i
hS||HSOL+
||Si 31.79 224.44i 32.07 228.23i
12/19
()<latexit sha1_base64="w7VdQLPr6UX/PCsJrACtfR9iswg=">AAAB8nicbVDLSgMxFM34rPVVdekmWITqosxUQZdFNy4r2AdMx5JJM21oJhmSO0IZ+hluXCji1q9x59+YtrPQ1kMCh3Pu5d57wkRwA6777aysrq1vbBa2its7u3v7pYPDllGppqxJlVC6ExLDBJesCRwE6ySakTgUrB2Obqd++4lpw5V8gHHCgpgMJI84JWAlv9JNuH2P52e4Vyq7VXcGvEy8nJRRjkav9NXtK5rGTAIVxBjfcxMIMqKBU8EmxW5qWELoiAyYb6kkMTNBNlt5gk+t0seR0vZLwDP1d0dGYmPGcWgrYwJDs+hNxf88P4XoOsi4TFJgks4HRanAoPD0ftznmlEQY0sI1dzuiumQaELBplS0IXiLJy+TVq3qXVRr95fl+k0eRwEdoxNUQR66QnV0hxqoiShS6Bm9ojcHnBfn3fmYl644ec8R+gPn8wfYDZBP</latexit>
(n)<latexit sha1_base64="bLd8oljV8VqhTI6XlWU5qwwy0NA=">AAAB8HicbVBNSwMxEJ2tX7V+VT16CRaheii7VdBj0YvHCvZD2rVk02wbmmSXJCuUpb/CiwdFvPpzvPlvTNs9aOuDgcd7M8zMC2LOtHHdbye3srq2vpHfLGxt7+zuFfcPmjpKFKENEvFItQOsKWeSNgwznLZjRbEIOG0Fo5up33qiSrNI3ptxTH2BB5KFjGBjpYey7Mbs8ewU9Yolt+LOgJaJl5ESZKj3il/dfkQSQaUhHGvd8dzY+ClWhhFOJ4VuommMyQgPaMdSiQXVfjo7eIJOrNJHYaRsSYNm6u+JFAutxyKwnQKboV70puJ/Xicx4ZWfMhknhkoyXxQmHJkITb9HfaYoMXxsCSaK2VsRGWKFibEZFWwI3uLLy6RZrXjnlerdRal2ncWRhyM4hjJ4cAk1uIU6NICAgGd4hTdHOS/Ou/Mxb8052cwh/IHz+QNZj490</latexit>
Analysis: Angular momentum
h1e(i) =X
K
ZK (ri RK) pi
|ri RK |3
Angular momentum around K
Rapid decay with distance
Real harmonics, the same center:
Lz =
0B@
py pz px
py 0 0 i
pz 0 0 0
px i 0 0
1CA L+ =
0B@
py pz px
py 0 i 0
pz i 0 1
px 0 1 0
1CA
El-Sayed rules: large SOC is achieved when the orientation ofp-orbitals changes, e.g. px ! pz.
9/19
Analysis: Angular momentum
h1e(i) =X
K
ZK (ri RK) pi
|ri RK |3
Angular momentum around K
Rapid decay with distance
Real harmonics, the same center:
Lz =
0B@
py pz px
py 0 0 i
pz 0 0 0
px i 0 0
1CA L+ =
0B@
py pz px
py 0 i 0
pz i 0 1
px 0 1 0
1CA
El-Sayed rules: large SOC is achieved when the orientation ofp-orbitals changes, e.g. px ! pz.
9/19
FIG. 12: El-Sayed’s rules for spin-forbidden transitions. Top: One-electron part of the Breit–Pauli
Hamiltonian has a form of local angular momentum on center K (marked by purple), owing to the
rapid decay with distance (denominator marked by pink). Matrix representation of Lz illustrates
that non-zero matrix elements involve the change of angular momentum (flip of the orbital). Top
right: By using this analysis, El-Sayed’s rules explain small magnitude of SOC for ππ∗ transitions
(no orbital flip) and large magnitude of SOC for nπ∗ transitions in organic molecules. Bottom:
NTOs of the spinless density matrix of Fe(II) single-molecule magnet explain large magnitude of
SOC and illustrate El-Sayed’s rules for d-orbitals. Reproduced with permission from Ref. 68.
D. Orbital concepts in non-Hermitian quantum chemistry
Standard quantum chemistry deals with electronically bound states (i.e., states stable
with respect to electron ejection). These states are discrete and their wavefunctions are
finite (L2-integrable). Ground states and low-lying excited states of many molecules are of
this type. Yet, many important physical phenomena take us outside the bound part of the
spectrum into the electronic continuum72–75. Examples include highly excited states (any
![Page 26: From orbitals to observables and backiopenshell.usc.edu/pubs/pdf/jcp_perspective.pdfQuantum mechanics explains chemical and physical phenomena by using abstract math-ematical constructs,](https://reader034.vdocuments.us/reader034/viewer/2022042406/5f1fa754210ddc366d060e82/html5/thumbnails/26.jpg)
26
excited state above the ionization threshold), core-level states, excited and sometimes even
ground states of anions (i.e., transient anions), and molecules on metal surfaces. In the
continuum, there are no discrete states and all wavefunctions are decaying asymptotically.
Yet, the continuum has a structure, revealed spectroscopically. It contains states, called
resonances, that behave as bound states, but with a finite lifetime. In Hermitian quantum
mechanics, these states correspond to an increased density of states in certain parts of the
continuum. Hence, standard quantum chemistry is not applicable—neither for a quantitative
treatment of these states nor for their conceptual analysis. The increased density of states
does not have a potential energy surface governing nuclear motion. The increased density of
states cannot be easily interpreted using molecular orbital theory. Yet, spectral signatures
of resonances, such as vibrational progressions and angular distributions of photoelectrons,
suggest that the concepts of Born–Oppenheimer separation of nuclear and electronic motion
and a molecular orbital picture of the transitions, do apply to these states hidden in the
continuum.
Non-Hermitian quantum mechanics72,73,75 offers an elegant framework for tackling reso-
nances. By recasting the Schrodinger equation into a non-Hermitian form, one can project
out the continuum and describe the resonances as isolated states with L2-integrable wave-
functions and complex energies. The real part of the resonance energy corresponds to the
resonance position and the imaginary part corresponds to the resonance width (inverse life-
time). By using complex potential energy surfaces76–79, one can describe nuclear dynamics
associated with resonances. Similar to bound states, observable properties of resonances are
encoded in the respective state and transition density matrices. Within the non-Hermitian
formalism, these densities become complex-valued80,81. For example, Dyson orbitals and
NTOs acquire real and imaginary components.
What can we learn from these complex-valued orbitals? In order to understand the
physical meaning of real and imaginary densities, one needs to relate them to experimental
observables. The lifetime of a resonance is the key observable. It is related to the imaginary
density, opening a route to interpretation. Further insight into the meaning of imaginary
densities and orbitals can be obtained by following the Feshbach-Fano formalism82. Within
this framework81, wavefunctions of metastable excited states (such as, for example, excited
states in many anions83) can be analyzed in terms of real and imaginary excitons via direct
extension of Eq. (12). Figure 13 shows such real and imaginary NTOs for two excited
![Page 27: From orbitals to observables and backiopenshell.usc.edu/pubs/pdf/jcp_perspective.pdfQuantum mechanics explains chemical and physical phenomena by using abstract math-ematical constructs,](https://reader034.vdocuments.us/reader034/viewer/2022042406/5f1fa754210ddc366d060e82/html5/thumbnails/27.jpg)
27
states of CN−. The real NTO pairs reveal that the two states can be described as π → π∗
and σ → π∗ excitations. The imaginary NTOs characterize the decay channels due to the
coupling to the continuum—the hole orbitals are very similar to the real hole orbitals, but
the particle orbitals have different nodal structure and are more diffuse. Figure 14 shows real
and imaginary NTOs for C3N− and C7N
−. The NTOs in the 1∆ state of C3N− are similar
to CN−, revealing π → π∗ character of the bound part of the exciton and a p-like decay
channel. In the 1Σ+ state of C7N−, the NTOs have more complex character: the bound
part of the exciton is a mixture of π → π∗ and σ → σ(Ry) excitation and the imaginary
part shows two different decay channels. Note that the relative weights of the configurations
involving the π and σ holes are flipped in the real and imaginary parts—in the real exciton,
the contribution of the σ hole is minor, whereas in the imaginary part it becomes the leading
component. Recall that, while the singular values for the real part have similar meaning as
in the bound-state calculations (they describe the character of the transition), the singular
values of the imaginary part are related to the partial widths of the resonance81. Hence,
this flip of leading σK reveals that this state of ππ∗ character decays predominantly via the
σσ(Ry) channel.
24
Analysis: Extended NTOs
Example: tris(pyrrolylmethyl)amine Fe(II) complex,quintet-quintet transition, EOM-EA-MP2
cm1 NTO integrals Reduced matrix elements (exact)
hS||HSOL ||Si 31.79 + 224.48i 32.07 + 228.23i
hS||HSOL0
||Si 0.04i 0.07i
hS||HSOL+
||Si 31.79 224.44i 32.07 228.23i
12/19
()<latexit sha1_base64="w7VdQLPr6UX/PCsJrACtfR9iswg=">AAAB8nicbVDLSgMxFM34rPVVdekmWITqosxUQZdFNy4r2AdMx5JJM21oJhmSO0IZ+hluXCji1q9x59+YtrPQ1kMCh3Pu5d57wkRwA6777aysrq1vbBa2its7u3v7pYPDllGppqxJlVC6ExLDBJesCRwE6ySakTgUrB2Obqd++4lpw5V8gHHCgpgMJI84JWAlv9JNuH2P52e4Vyq7VXcGvEy8nJRRjkav9NXtK5rGTAIVxBjfcxMIMqKBU8EmxW5qWELoiAyYb6kkMTNBNlt5gk+t0seR0vZLwDP1d0dGYmPGcWgrYwJDs+hNxf88P4XoOsi4TFJgks4HRanAoPD0ftznmlEQY0sI1dzuiumQaELBplS0IXiLJy+TVq3qXVRr95fl+k0eRwEdoxNUQR66QnV0hxqoiShS6Bm9ojcHnBfn3fmYl644ec8R+gPn8wfYDZBP</latexit>
(n)<latexit sha1_base64="bLd8oljV8VqhTI6XlWU5qwwy0NA=">AAAB8HicbVBNSwMxEJ2tX7V+VT16CRaheii7VdBj0YvHCvZD2rVk02wbmmSXJCuUpb/CiwdFvPpzvPlvTNs9aOuDgcd7M8zMC2LOtHHdbye3srq2vpHfLGxt7+zuFfcPmjpKFKENEvFItQOsKWeSNgwznLZjRbEIOG0Fo5up33qiSrNI3ptxTH2BB5KFjGBjpYey7Mbs8ewU9Yolt+LOgJaJl5ESZKj3il/dfkQSQaUhHGvd8dzY+ClWhhFOJ4VuommMyQgPaMdSiQXVfjo7eIJOrNJHYaRsSYNm6u+JFAutxyKwnQKboV70puJ/Xicx4ZWfMhknhkoyXxQmHJkITb9HfaYoMXxsCSaK2VsRGWKFibEZFWwI3uLLy6RZrXjnlerdRal2ncWRhyM4hjJ4cAk1uIU6NICAgGd4hTdHOS/Ou/Mxb8052cwh/IHz+QNZj490</latexit>
Analysis: Angular momentum
h1e(i) =X
K
ZK (ri RK) pi
|ri RK |3
Angular momentum around K
Rapid decay with distance
Real harmonics, the same center:
Lz =
py pz px
py 0 0 i
pz 0 0 0
px i 0 0
L+ =
py pz px
py 0 i 0
pz i 0 1
px 0 1 0
El-Sayed rules: large SOC is achieved when the orientation ofp-orbitals changes, e.g. px ! pz.
9/19
Analysis: Angular momentum
h1e(i) =X
K
ZK (ri RK) pi
|ri RK |3
Angular momentum around K
Rapid decay with distance
Real harmonics, the same center:
Lz =
py pz px
py 0 0 i
pz 0 0 0
px i 0 0
L+ =
py pz px
py 0 i 0
pz i 0 1
px 0 1 0
El-Sayed rules: large SOC is achieved when the orientation ofp-orbitals changes, e.g. px ! pz.
9/19
FIG. 12: El-Sayed rules for spin-forbidden transitions. Top: One-electron part of the Breight-Pauli
Hamiltonian has a form of local angular momentum on center K (marked by purple), owing to the
rapid decay with distance (denominator marked by pink). Matrix representation of Lz isssutrates
that non-zero matrix elements involve the change of anumar moemnum (flip of the orbital). Top
right: By using this analysis, El-Sayed rules expain small magntitude if SOC for transitions
(no orbital flip) and large magnitude of SOC for n transitions in organic molecles. Bottom:
NTOs of the spinless density matrix of Fe(II) SMM explain large magnitude of SOC and illustrate
El-Sayed rules for d-orbitals.
Orbital characterization of the excited states
Resonances correspond to valence transition of the type (+, , ) or ()
1+ resonance in C7N is an exception – it has a strongly mixed and characterW. Skomorowski Real and imaginary excitons 9 / 14
Redo diagrams
FIG. 13: Real and imaginary NTOs in the two excited states of CN.
24
Analysis: Extended NTOs
Example: tris(pyrrolylmethyl)amine Fe(II) complex,quintet-quintet transition, EOM-EA-MP2
cm1 NTO integrals Reduced matrix elements (exact)
hS||HSOL ||Si 31.79 + 224.48i 32.07 + 228.23i
hS||HSOL0
||Si 0.04i 0.07i
hS||HSOL+
||Si 31.79 224.44i 32.07 228.23i
12/19
()<latexit sha1_base64="w7VdQLPr6UX/PCsJrACtfR9iswg=">AAAB8nicbVDLSgMxFM34rPVVdekmWITqosxUQZdFNy4r2AdMx5JJM21oJhmSO0IZ+hluXCji1q9x59+YtrPQ1kMCh3Pu5d57wkRwA6777aysrq1vbBa2its7u3v7pYPDllGppqxJlVC6ExLDBJesCRwE6ySakTgUrB2Obqd++4lpw5V8gHHCgpgMJI84JWAlv9JNuH2P52e4Vyq7VXcGvEy8nJRRjkav9NXtK5rGTAIVxBjfcxMIMqKBU8EmxW5qWELoiAyYb6kkMTNBNlt5gk+t0seR0vZLwDP1d0dGYmPGcWgrYwJDs+hNxf88P4XoOsi4TFJgks4HRanAoPD0ftznmlEQY0sI1dzuiumQaELBplS0IXiLJy+TVq3qXVRr95fl+k0eRwEdoxNUQR66QnV0hxqoiShS6Bm9ojcHnBfn3fmYl644ec8R+gPn8wfYDZBP</latexit>
(n)<latexit sha1_base64="bLd8oljV8VqhTI6XlWU5qwwy0NA=">AAAB8HicbVBNSwMxEJ2tX7V+VT16CRaheii7VdBj0YvHCvZD2rVk02wbmmSXJCuUpb/CiwdFvPpzvPlvTNs9aOuDgcd7M8zMC2LOtHHdbye3srq2vpHfLGxt7+zuFfcPmjpKFKENEvFItQOsKWeSNgwznLZjRbEIOG0Fo5up33qiSrNI3ptxTH2BB5KFjGBjpYey7Mbs8ewU9Yolt+LOgJaJl5ESZKj3il/dfkQSQaUhHGvd8dzY+ClWhhFOJ4VuommMyQgPaMdSiQXVfjo7eIJOrNJHYaRsSYNm6u+JFAutxyKwnQKboV70puJ/Xicx4ZWfMhknhkoyXxQmHJkITb9HfaYoMXxsCSaK2VsRGWKFibEZFWwI3uLLy6RZrXjnlerdRal2ncWRhyM4hjJ4cAk1uIU6NICAgGd4hTdHOS/Ou/Mxb8052cwh/IHz+QNZj490</latexit>
Analysis: Angular momentum
h1e(i) =X
K
ZK (ri RK) pi
|ri RK |3
Angular momentum around K
Rapid decay with distance
Real harmonics, the same center:
Lz =
py pz px
py 0 0 i
pz 0 0 0
px i 0 0
L+ =
py pz px
py 0 i 0
pz i 0 1
px 0 1 0
El-Sayed rules: large SOC is achieved when the orientation ofp-orbitals changes, e.g. px ! pz.
9/19
Analysis: Angular momentum
h1e(i) =X
K
ZK (ri RK) pi
|ri RK |3
Angular momentum around K
Rapid decay with distance
Real harmonics, the same center:
Lz =
py pz px
py 0 0 i
pz 0 0 0
px i 0 0
L+ =
py pz px
py 0 i 0
pz i 0 1
px 0 1 0
El-Sayed rules: large SOC is achieved when the orientation ofp-orbitals changes, e.g. px ! pz.
9/19
FIG. 12: El-Sayed rules for spin-forbidden transitions. Top: One-electron part of the Breight-Pauli
Hamiltonian has a form of local angular momentum on center K (marked by purple), owing to the
rapid decay with distance (denominator marked by pink). Matrix representation of Lz isssutrates
that non-zero matrix elements involve the change of anumar moemnum (flip of the orbital). Top
right: By using this analysis, El-Sayed rules expain small magntitude if SOC for transitions
(no orbital flip) and large magnitude of SOC for n transitions in organic molecles. Bottom:
NTOs of the spinless density matrix of Fe(II) SMM explain large magnitude of SOC and illustrate
El-Sayed rules for d-orbitals.
Orbital characterization of the excited states
Resonances correspond to valence transition of the type (+, , ) or ()
1+ resonance in C7N is an exception – it has a strongly mixed and characterW. Skomorowski Real and imaginary excitons 9 / 14
Redo diagrams
FIG. 13: Real and imaginary NTOs in the two excited states of CN.
<latexit sha1_base64="MMfoOZUbGzRkaB76umvTEWj+CN8=">AAAB6nicbVBNS8NAEJ3Ur1q/qh69LBbBU0mqoMeiF48V7Qe0oWy2k3bpZhN2N0IJ/QlePCji1V/kzX/jts1BWx8MPN6bYWZekAiujet+O4W19Y3NreJ2aWd3b/+gfHjU0nGqGDZZLGLVCahGwSU2DTcCO4lCGgUC28H4dua3n1BpHstHM0nQj+hQ8pAzaqz00Et4v1xxq+4cZJV4OalAjka//NUbxCyNUBomqNZdz02Mn1FlOBM4LfVSjQllYzrErqWSRqj9bH7qlJxZZUDCWNmShszV3xMZjbSeRIHtjKgZ6WVvJv7ndVMTXvsZl0lqULLFojAVxMRk9jcZcIXMiIkllClubyVsRBVlxqZTsiF4yy+vklat6l1Ua/eXlfpNHkcRTuAUzsGDK6jDHTSgCQyG8Ayv8OYI58V5dz4WrQUnnzmGP3A+fwBRC43R</latexit>
<latexit sha1_base64="UyKAkdDl8Y1enA6pUASuPNFQvcY=">AAAB7HicbVBNS8NAEJ34WetX1aOXxSKIh5JUQY9FLx4rmLbQxrLZbtqlm03YnQil9Dd48aCIV3+QN/+N2zYHbX0w8Hhvhpl5YSqFQdf9dlZW19Y3Ngtbxe2d3b390sFhwySZZtxniUx0K6SGS6G4jwIlb6Wa0ziUvBkOb6d+84lrIxL1gKOUBzHtKxEJRtFKficVj+fdUtmtuDOQZeLlpAw56t3SV6eXsCzmCpmkxrQ9N8VgTDUKJvmk2MkMTykb0j5vW6pozE0wnh07IadW6ZEo0bYUkpn6e2JMY2NGcWg7Y4oDs+hNxf+8dobRdTAWKs2QKzZfFGWSYEKmn5Oe0JyhHFlCmRb2VsIGVFOGNp+iDcFbfHmZNKoV76JSvb8s127yOApwDCdwBh5cQQ3uoA4+MBDwDK/w5ijnxXl3PuatK04+cwR/4Hz+AGvUjm0=</latexit>
<latexit sha1_base64="uveq51XskeZ/BmBbyb/DEzkG8yU=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKexGQY9BLx4jmAckS5idzCZj5rHMzAphyT948aCIV//Hm3/jJNmDJhY0FFXddHdFCWfG+v63V1hb39jcKm6Xdnb39g/Kh0cto1JNaJMornQnwoZyJmnTMstpJ9EUi4jTdjS+nfntJ6oNU/LBThIaCjyULGYEWye1eoYNBe6XK37VnwOtkiAnFcjR6Je/egNFUkGlJRwb0w38xIYZ1pYRTqelXmpogskYD2nXUYkFNWE2v3aKzpwyQLHSrqRFc/X3RIaFMRMRuU6B7cgsezPxP6+b2vg6zJhMUkslWSyKU46sQrPX0YBpSiyfOIKJZu5WREZYY2JdQCUXQrD88ipp1arBRbV2f1mp3+RxFOEETuEcAriCOtxBA5pA4BGe4RXePOW9eO/ex6K14OUzx/AH3ucPna2PJw==</latexit>
<latexit sha1_base64="MMfoOZUbGzRkaB76umvTEWj+CN8=">AAAB6nicbVBNS8NAEJ3Ur1q/qh69LBbBU0mqoMeiF48V7Qe0oWy2k3bpZhN2N0IJ/QlePCji1V/kzX/jts1BWx8MPN6bYWZekAiujet+O4W19Y3NreJ2aWd3b/+gfHjU0nGqGDZZLGLVCahGwSU2DTcCO4lCGgUC28H4dua3n1BpHstHM0nQj+hQ8pAzaqz00Et4v1xxq+4cZJV4OalAjka//NUbxCyNUBomqNZdz02Mn1FlOBM4LfVSjQllYzrErqWSRqj9bH7qlJxZZUDCWNmShszV3xMZjbSeRIHtjKgZ6WVvJv7ndVMTXvsZl0lqULLFojAVxMRk9jcZcIXMiIkllClubyVsRBVlxqZTsiF4yy+vklat6l1Ua/eXlfpNHkcRTuAUzsGDK6jDHTSgCQyG8Ayv8OYI58V5dz4WrQUnnzmGP3A+fwBRC43R</latexit>
<latexit sha1_base64="UyKAkdDl8Y1enA6pUASuPNFQvcY=">AAAB7HicbVBNS8NAEJ34WetX1aOXxSKIh5JUQY9FLx4rmLbQxrLZbtqlm03YnQil9Dd48aCIV3+QN/+N2zYHbX0w8Hhvhpl5YSqFQdf9dlZW19Y3Ngtbxe2d3b390sFhwySZZtxniUx0K6SGS6G4jwIlb6Wa0ziUvBkOb6d+84lrIxL1gKOUBzHtKxEJRtFKficVj+fdUtmtuDOQZeLlpAw56t3SV6eXsCzmCpmkxrQ9N8VgTDUKJvmk2MkMTykb0j5vW6pozE0wnh07IadW6ZEo0bYUkpn6e2JMY2NGcWg7Y4oDs+hNxf+8dobRdTAWKs2QKzZfFGWSYEKmn5Oe0JyhHFlCmRb2VsIGVFOGNp+iDcFbfHmZNKoV76JSvb8s127yOApwDCdwBh5cQQ3uoA4+MBDwDK/w5ijnxXl3PuatK04+cwR/4Hz+AGvUjm0=</latexit>
<latexit sha1_base64="uveq51XskeZ/BmBbyb/DEzkG8yU=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKexGQY9BLx4jmAckS5idzCZj5rHMzAphyT948aCIV//Hm3/jJNmDJhY0FFXddHdFCWfG+v63V1hb39jcKm6Xdnb39g/Kh0cto1JNaJMornQnwoZyJmnTMstpJ9EUi4jTdjS+nfntJ6oNU/LBThIaCjyULGYEWye1eoYNBe6XK37VnwOtkiAnFcjR6Je/egNFUkGlJRwb0w38xIYZ1pYRTqelXmpogskYD2nXUYkFNWE2v3aKzpwyQLHSrqRFc/X3RIaFMRMRuU6B7cgsezPxP6+b2vg6zJhMUkslWSyKU46sQrPX0YBpSiyfOIKJZu5WREZYY2JdQCUXQrD88ipp1arBRbV2f1mp3+RxFOEETuEcAriCOtxBA5pA4BGe4RXePOW9eO/ex6K14OUzx/AH3ucPna2PJw==</latexit>
FIG. 13: Real and imaginary NTOs for the Π (top) and Σ+/∆/Σ resonances in CN−. CAP-EOM-
EE-CCSD.
![Page 28: From orbitals to observables and backiopenshell.usc.edu/pubs/pdf/jcp_perspective.pdfQuantum mechanics explains chemical and physical phenomena by using abstract math-ematical constructs,](https://reader034.vdocuments.us/reader034/viewer/2022042406/5f1fa754210ddc366d060e82/html5/thumbnails/28.jpg)
28
C3N-
C7N-
1
Partial widths: Γ(Σ) = 0.10 eVΓ(Π) = 0.03 eVΓtot = 0.13 eV
14
Real
Real
Real
Imaginary
Imaginary
Imaginary
! = #. %&
! = #. %&
! = #. '(
! = #. )(
! = #. #%
! = #. #%
FIG. 3: Real and imaginary NTOs and their corresponding singular values K for the 1+ resonance
state in C7N.
Figure 4 shows average hole sizes for the valence triplet and singlet states in C2n+1N,
n = 0, . . . , 3. As one can see, in the states the hole size remains nearly the same (or
even shrinks slightly), because orbital is localized, whereas in and + states it increases
with n, indicating the delocalized nature of the orbital. The shapes and sizes of real
and imaginary hole NTOs are similar. In contrast, the sizes of the real and imaginary
particle orbitals are di↵erent (see Fig. S2 in SI), the latter being more di↵use because of
their continuum origin. These di↵erent trends in sizes of holes and particles are also clearly
reflected in the average hole-particle separation, as shown in Fig. 5. Imaginary excitons
are characterized by significantly larger hole-particle distance than the real excitons, which
illustrates our point of Im enhancing the decay channels via the virtual states.
We also analyze electron-hole correlation, Reh, (shown in Fig. S3 in SI). As expected for
molecular excitonic states, the real excitons show positive correlation, indicative of electron-
hole attraction resulting in a correlated motion of a hole and a particle. The magnitude of
Reh(Re) is similar for all states and is close to the values in Mg porphyrin reported recently in
Ref. 33. Interestingly, the imaginary excitons show much smaller correlation and sometimes
even negative values, meaning that an electron and a hole move independently (or away
from each other) and that the imaginary exciton is unbound. This is again consistent with
the interpretation of the imaginary part of the exciton as a virtual decay channel.
Finally, Fig. 6 shows the correlation between the magnitude of the imaginary part of
the exciton with the resonance width. Indeed, as suggested by Eqs. (30) and (31), there
13
in Fig. 3, we estimate = 0.13 0.142
0.142+2·0.052 = 0.10 eV and = 0.13 2·0.0.052
0.142+2·0.052 = 0.03 eV.
All other resonances identified in cyanopolyyne anions8 are of shape type decaying almost
exclusively through a single decay channel, which is reflected by the respective NTOs and
their singular values.
In addition to these valence excited states, some cyanopolyynes also support dipole-
stabilized states. Figure S1 in SI shows NTOs for the 3+ dipole-stabilized resonance of
C5N. The shapes of NTOs clearly reveal the di↵erent nature of this state. The imaginary
part of the exciton has a d-wave character.
Real
Imaginary
Real
Real Real
Imaginary
CN- C3N-
C5N- C7N-
! = #. %&
! = #. '&
! = #. %(
! = #. #&
! = #. )% ! = #. )%
FIG. 1: Real and imaginary NTOs and their corresponding singular values K for the 3 state in
four cyanopolyyne anions: CN, C3N, C5N
, and C7N.
Real
Real
Real
Real
Imaginary
Imaginary
Imaginary
Imaginary
CN-
C3N-
! = #. %%
! = #. %%
! = #. &'
! = #. &'
! = #. &(
! = #. &(
! = #. #)
! = #. #)
FIG. 2: Real and imaginary NTOs and their corresponding singular values K for the 1 state in
CN and C3N.
1+
FIG. 14: Real and imaginary NTOs in C3N− (top) and C7N
− (bottom); CAP-EOM-EE-CCSD.
Reprinted with permission from Ref. 81. Copyright (2018) American Chemical Society..
![Page 29: From orbitals to observables and backiopenshell.usc.edu/pubs/pdf/jcp_perspective.pdfQuantum mechanics explains chemical and physical phenomena by using abstract math-ematical constructs,](https://reader034.vdocuments.us/reader034/viewer/2022042406/5f1fa754210ddc366d060e82/html5/thumbnails/29.jpg)
29
III. CAN WE REALLY SEE THEM?
LETTERS NATURE PHYSICS DOI: 10.1038/NPHYS3212
−0.05 V−0.26 V−0.72 V
−0.15 V−0.35 V−0.81 V
dI/d
V
Hig
hLo
w
−1.0 −0.5 0.0 0.50
1
2
3a
b
c
d
A4 A3 A2
A4v A3v A2v
−1.0 −0.5 0.00
1
2
3
dI/d
V (a
.u.)
E − E0N (eV)
LDO
S (a
.u.)
A2
A2v
A3A3v
A4A4v
Sample bias (V)
CoPC−1
ψ⎜ 2N ⟩
ψ⎜ 0N ⟩
ψ⎜ 2N ⟩ ψ⎜ 1
N ⟩ ψ⎜ 0N ⟩ ψ⎜ 0
N +1⟩
ψ⎜ 1N ⟩
Figure 3 | Wavefunction mapping on the negatively charged CoPC. a,b, Experimental high-resolution dI/dV spectra (a) and simulated LDOS (b) ondifferent points over the CoPC−1. Spectra are offset for clarity. The hump at ∼−0.65V in the blue spectrum in a was not reproducibly found on othermolecules (see Supplementary Information) and thus not assigned to a molecular resonance. c, Experimental dI/dV maps at different energies,corresponding to the different resonances (middle row) and their vibronic replica (upper row). The lower row shows simulated LDOS maps for the MBtransitions from CoPC−1 to CoPC0. d, STM image overlaid with the structural model of CoPC, indicating the positions of the experimental and simulatedspectra, respectively. Scale bar is 1 nm, tunnelling parameters are −1.51 V, 8 pA.
0.5 1.0 1.5 2.0
0.0 0.5 1.00
1
2
3 −1.01 VN3 1.93 VN1vv1.69 VN10.98 VN2
1.79 VN1v1.08 VN2v
dI/d
V
Hig
hLo
w
N2
N2v
N1
N1vN1vv
Sample bias (V)
E − E0N+1 (eV)
LDO
S (a
.u.)
0
1
2
3a c d
b
dI/d
V (a
.u.)
CoPC0
ψ⎜ 0N+1 ⟩
ψ⎜ 2N+1 ⟩
ψ⎜ 5N+1 ⟩
ψ⎜ 2N+1 ⟩ ψ⎜ 5
N+1 ⟩ψ⎜ 0N−1 ⟩ ψ⎜ 0
N ⟩ ψ⎜ 0N ⟩ ψ⎜ 0
N+1 ⟩
Figure 4 | Wavefunction mapping on the neutral CoPC. a,b, Experimental high-resolution dI/dV spectra (a) and simulated LDOS (b) on different pointsover the CoPC0. Spectra are offset for clarity. c, Experimental dI/dV maps at different energies, corresponding to the different resonances (middle row)and their vibronic replica (upper row). The lower row shows simulated LDOS maps for the MB transitions from CoPC0 to CoPC−1, as well as from CoPC0
into the CoPC+1 ground state at negative energy. d, STM image overlaid with the structural model of CoPC, indicating the positions of the experimental andsimulated spectra, respectively. Scale bar is 1 nm, tunnelling parameters are −2.00V, 2 pA.
of the states from which or into which the tunnelling occurs.The dI/dV spectra taken on different positions over the CoPCmolecule can be simulated by multiplying the total DOS shown inFig. 2 by the squares of the relevant wavefunctions at the positionof the STM tip. Figure 3a shows experimental high-resolutionspectra taken on three different positions over CoPC−1. Clearly, therelative intensities of the A2, A3 and A4 resonances change withposition. This is in good agreement with the simulated spectra for
the ground state |Ψ N0 ⟩ and the MB excited states |Ψ N
1 ⟩ and |Ψ N2 ⟩
(Fig. 3b), and directly reflects the different symmetries and nodesof the wavefunctions involved in the transitions (see SupplementaryInformation for simulated SP spectra). Their shape and symmetrycan be directly visualized by recording the dI/dV signal in theconstant-height mode, allowing comparison with simulated LDOSmaps (Fig. 3c). The wavefunctions of the excited states |Ψ N
1 ⟩ and|Ψ N
2 ⟩ both have four-fold symmetry but different nodal planes.
232 NATURE PHYSICS | VOL 11 | MARCH 2015 | www.nature.com/naturephysics
© 2015 Macmillan Publishers Limited. All rights reserved
FIG. 15: STM image of neutral cobalt phthalocyanine (CoPC) overlaid with the molecular struc-
ture. Experimental differential conductance maps at different energies, corresponding to the differ-
ent resonances (middle row) and their vibronic replica (upper row). The lower row shows simulated
local density of states maps for the transitions from CoPC0 to CoPC1 and from CoPC0 to CoPC+1,
representing squares of the Dyson orbitals of the electron-attached and ionized states, respectively.
Reproduced with permission from Ref. 84.
The theory provides a solid foundation for experimental orbital imaging, but design
and implementation of such experiments are far from trivial. The interpretation of orbital
imaging experiments is still controversial, highlighting the need for more robust theoretical
modeling and data processing frameworks. Despite these challenges, there are many exciting
examples illustrating our progress towards direct experimental orbital imaging.
STM enables atomic-scale imaging of molecules adsorbed on surfaces. By measuring
changes in the local molecular density of states under different applied voltages, STM can
provide real-space images of the electron density of the Dyson orbitals that connect the
neutral molecules with their respective negative and positive ions. Figure 15 shows images
for cobalt phthalocyanine (CoPC). The experimental differential conductance maps reveal
clear nodal structure of the Dyson orbitals, in good agreement with the computed maps.
However, the phase information is not available and, at present, STM imaging is limited to
the two-dimensional picture. Additional complications involve perturbation of the electronic
structure by the STM tip and the surface.
![Page 30: From orbitals to observables and backiopenshell.usc.edu/pubs/pdf/jcp_perspective.pdfQuantum mechanics explains chemical and physical phenomena by using abstract math-ematical constructs,](https://reader034.vdocuments.us/reader034/viewer/2022042406/5f1fa754210ddc366d060e82/html5/thumbnails/30.jpg)
30
Advanced light sources offer new possibilities for orbital imaging, such as the use of high-
order harmonic generation85. An addition, strong fields can be used to align molecules in
space, which eliminates orientational averaging and, therefore, increases the amount of in-
formation that can be used for reconstruction. Angular-resolved photoelectron experiments,
especially when executed for oriented molecules, can deliver three-dimensional images of
orbitals86–89, however, the exact details of the reconstruction process remain a key chal-
lenge. Given that the interpretation of these experiments often involves certain simplifying
assumptions (such as a plane-wave treatment of photoelectrons) and multiple steps of con-
verting the raw data into the image, one may argue that we are not yet truly able to see
the orbitals, at least not in a way that would convince Mach and his fellow members of the
Vienna Circle. Challenges of the full tomographic orbital imaging aside, the mapping be-
tween orbitals and observables is indisputable, and this connection provides a sound guiding
principle for responsible use of orbital concepts in physics and chemistry.
IV. CONCLUSION
In this contribution, I have discussed molecular orbital concepts from the point of view
of many-body electronic structure and observable molecular properties. The examples il-
lustrate how simple molecular orbital pictures can be rigorously distilled from correlated
wavefunctions by using reduced quantities such as one-particle transition density matrices.
By using this formalism, molecular orbitals can be related to molecular properties and spec-
troscopic observables. Thus, orbitals are not a mere mathematical abstraction; they are
real and can be observed. However, the type of orbitals you see depends on which tool you
use to look at them. This is an intrinsic feature of quantum mechanics — the result of the
observation depends not only on the state of the system but also on the act of observation.
Orbital concepts are instrumental in deriving physical insight from spectroscopy. Here again
different spectroscopies relate to different types of orbitals. I conclude with a general guiding
principle for using molecular orbital concepts: Follow the observable. Or design a new one!
![Page 31: From orbitals to observables and backiopenshell.usc.edu/pubs/pdf/jcp_perspective.pdfQuantum mechanics explains chemical and physical phenomena by using abstract math-ematical constructs,](https://reader034.vdocuments.us/reader034/viewer/2022042406/5f1fa754210ddc366d060e82/html5/thumbnails/31.jpg)
31
Acknowledgments
I wish to dedicate this work to Dr. Vladimir Ivanovich Pupyshev (08/06/1949-
06/16/2020), who passed away90 as this manuscript was being finalized. Dr. Pupyshev was
an inspiring teacher and a gentle mentor, who ignited my interest in quantum chemistry
and encouraged my initial clumsy attempts in theoretical work. Dr. Pupyshev was able to
convey the conceptual beauty of quantum mechanics while maintaining the highest level of
mathematical rigor, always spiced with a good joke, a literary quote, or a thought-provoking
sharp remark. His scientific style has provided a constant inspiration for my work.
This work was supported by the U.S. National Science Foundation (No. CHE-1856342).
Conflicts of interest
A.I.K. is the president and a part-owner of Q-Chem, Inc.
Data availability statement
Data sharing is not applicable to this article as no new data were created or analyzed in
this work.
1 https://en.wikipedia.org/wiki/Ernst Mach.
2 People of my generation may find amusing the connection between Mach and Lenin. In his essay
”Materialism and Empirio-criticism”, Lenin scorned Mach and his followers, along with several
other ”reactionist philosophers”. Despite having extensive schooling in dialectical materialism
and Lenin’s work—we had to write extensive conspectuses of this particular treatise to pass
exams—I am completely blank on which part of Mach’s teaching Lenin deemed to be at odds
with Marxist’s philosophy. Its low intellectual content and poor scholarly quality notwithstand-
ing, ”Materialism and Empirio-criticism” impacted science in a profound way. The arrogance of
a poorly educated revolutionary zealot attacking competent scientists heralded the beginning
of a new era in which ideology trumps the science and politicians censor it. In the spirit of
”Materialism and Empirio-criticism”, Soviet commissars declared entire scientific disciplines as
![Page 32: From orbitals to observables and backiopenshell.usc.edu/pubs/pdf/jcp_perspective.pdfQuantum mechanics explains chemical and physical phenomena by using abstract math-ematical constructs,](https://reader034.vdocuments.us/reader034/viewer/2022042406/5f1fa754210ddc366d060e82/html5/thumbnails/32.jpg)
32
ideologically impure (reactionary, renegade, bourgeoisie, imperialistic, colonial, nationalistic,
etc) and antagonistic to the cause of the superior class (proletariat, in Soviet Russia), shutting
down research and sanctioning (up to physically eliminating) the researchers.
3 Royal Swedish Academy of Sciencies, Press Release, 1981, ”The Royal Swedish Academy of
Sciences has decided to award the 1981 Nobel Prize in chemistry jointly to Professor Kenichi
Fukui, Kyoto University, Kyoto, Japan, and the other half to Professor Roald Hoffmann, Cornell
University, Ithaca, NY, USA, for their theories, developed independently, concerning the course
of chemical reactions.”.
4 K. Ruedenberg, The physical nature of the chemical bond, Rev. Mod. Phys. 34, 326 (1962).
5 M. W. Schmidt, J. Ivanic, and K. Ruedenberg, Covalent bonds are created by the drive of
electron waves to lower their kinetic energy through expansion, J. Chem. Phys. 140, 204104
(2014).
6 M. P. Minitti, J. M. Budarz, A. Kirrander, J. S. Robinson, D. Ratner, T. J. Lane, D. Zhu,
J. M. Glownia, M. Kozina, H. T. Lemke, M. Sikorski, Y. Feng, S. Nelson, K. Saita, B. Stankus,
T. Northey, J. B. Hastings, and P. M. Weber, Imaging molecular motion: Femtosecond x-ray
scattering of an electrocyclic chemical reaction, Phys. Rev. Lett. 114, 255501 (2015).
7 H. Jean-Ruel, M. Gao, M. A. Kochman, C. Lu, L. C. Liu, R. R. Cooney, C. A. Morrison,
and R. J. D. Miller, Ring-closing reaction in diarylethene captured by femtosecond electron
crystallography, J. Phys. Chem. B 117, 15894 (2013).
8 T. Helgaker, P. Jørgensen, and J. Olsen, Molecular electronic structure theory. Wiley & Sons,
2000.
9 H. A. Bethe and E. E. Salpeter, Quantum mechanics of one and two electron atoms. Plenum,
New York, 1977.
10 The theory of photoionization can be formulated using the momentum gauge, leading to the
following expression for the photoelectron matrix element:
Dk = iu〈φd(r)|∇r|Ψelk (r)〉. (30)
.
11 C. M. Oana and A. I. Krylov, Dyson orbitals for ionization from the ground and electronically
excited states within equation-of-motion coupled-cluster formalism: Theory, implementation,
and examples, J. Chem. Phys. 127, 234106 (2007).
![Page 33: From orbitals to observables and backiopenshell.usc.edu/pubs/pdf/jcp_perspective.pdfQuantum mechanics explains chemical and physical phenomena by using abstract math-ematical constructs,](https://reader034.vdocuments.us/reader034/viewer/2022042406/5f1fa754210ddc366d060e82/html5/thumbnails/33.jpg)
33
12 C. M. Oana and A. I. Krylov, Cross sections and photoelectron angular distributions in pho-
todetachment from negative ions using equation-of-motion coupled-cluster Dyson orbitals, J.
Chem. Phys. 131, 124114 (2009).
13 S. Gozem and A. I. Krylov, ezDyson user’s manual, 2015,
ezDyson, http://iopenshell.usc.edu/downloads/. Accessed April 2018.
14 S. Gozem, A. O. Gunina, T. Ichino, D. L. Osborn, J. F. Stanton, and A. I. Krylov, Photoelectron
wave function in photoionization: Plane wave or Coulomb wave?, J. Phys. Chem. Lett. 6, 4532
(2015).
15 O. Goscinski and P. Lindner, Natural spin-orbitals and generalized overlap amplitudes, J.
Math. Phys. 11, 1313 (1970).
16 J. Linderberg and Y. Ohrn, Propagators in quantum chemistry. Academic, London, 1973.
17 R. L. Martin and E. R. Davidson, Halogen atomic and diatomic 1s hole states, Phys. Rev. A
16, 1341 (1977).
18 L.S. Cederbaum and W. Domcke, Theoretical aspects of ionization potentials and photoelectron
spectroscopy: A Green’s function approach, Adv. Chem. Phys. 36, 205 (1977).
19 J. V. Ortiz, Toward an exact one-electron picture of chemical bonding, Adv. Quantum Chem.
35, 33 (1999).
20 H. R. Hudock, B. G. Levine, A. L. Thompson, H. Satzger, D. Townsend, N. Gador, S. Ulrich,
A. Stolow, and T. J. Martınez, Ab initio molecular dynamics and time-resolved photoelectron
spectroscopy of electronically excited uracil and thymine, J. Phys. Chem. A 111, 8500 (2007).
21 When computed using the Hartree-Fock wave function of the neutral and neglecting orbital
relaxation in the cation, Dyson orbitals are canonical Hartree–Fock orbitals, as stipulated by
Koopmans’ theorem. If orbitals are allowed to relax, as in the ∆SCF treatment of ionization,
Dyson orbitals (as defined by Eq. (5)) can be computed by singular value decomposition of
the overlap matrix between the orbitals of the neutral and the ionized systems, as was done
by Martin and Davidson17 and described formally by Goscinski and Lindner15. Ref. 17, which
represents one of the earlier appearances of Dyson orbitals in mainstream quantum chemistry17,
used the term ”corresponding ionization orbitals” and did not make an explicit connection with
the Dyson orbitals (also called generalized overlap amplitudes) appearing in Green’s function
theory15,16. Rather, the theory was described in terms of the ”corresponding orbital transfor-
mation of Amos and Hall”91.
![Page 34: From orbitals to observables and backiopenshell.usc.edu/pubs/pdf/jcp_perspective.pdfQuantum mechanics explains chemical and physical phenomena by using abstract math-ematical constructs,](https://reader034.vdocuments.us/reader034/viewer/2022042406/5f1fa754210ddc366d060e82/html5/thumbnails/34.jpg)
34
22 I. E. McCarthy and E. Weigold, Wavefunction mapping in collision experiments, Rep. Prog.
Phys. 51, 299 (1988).
23 E. Weigold and I. E. McCarthy, Electron momentum spectroscopy. Kluwer, Dordecht/Plenum,
New York, 1999.
24 A. I. Krylov, Equation-of-motion coupled-cluster methods for open-shell and electronically
excited species: The hitchhiker’s guide to Fock space, Annu. Rev. Phys. Chem. 59, 433 (2008).
25 K. Sneskov and O. Christiansen, Excited state coupled cluster methods, WIREs: Comput.
Mol. Sci. 2, 566 (2012).
26 R. J. Bartlett, Coupled-cluster theory and its equation-of-motion extensions, WIREs: Comput.
Mol. Sci. 2, 126 (2012).
27 R. Mabbs, E. R. Grumbling, K. Pichugin, and A. Sanov, Photoelectron imaging: an experi-
mental window into electronic structure, Chem. Soc. Rev. 38, 2169 (2009).
28 A. Sanov and R. Mabbs, Photoelectron imaging of negative ions, Int. Rev. Phys. Chem. 27, 53
(2008).
29 A. Sanov, Photodetachment: Insight into electronic structure and intermolecular interactions,
Annu. Rev. Phys. Chem. 65, 341 (2014).
30 S. Gozem, R. Seidel, U. Hergenhahn, E. Lugovoy, B. Abel, B. Winter, A. I. Krylov, and S. E.
Bradfort, Probing the electronic structure of bulk water at the molecular length scale with
angle-resolved photoelectron spectroscopy, J. Phys. Chem. Lett. 11, 5162 (2020).
31 R. Mabbs, K. Pichugin, and A. Sanov, Dynamic molecular interferometer: probe of inversion
symmetry in I−2 photodissociation, J. Chem. Phys. 123, 054329 (2005).
32 M. V. Ivanov, F. H. Bangerter, and A. I. Krylov, Towards a rational design of laser-coolable
molecules: Insights from equation-of-motion coupled-cluster calculations, Phys. Chem. Chem.
Phys. 21, 19447 (2019).
33 I. Kozyryev, L. Baum, K. Matsuda, B. L. Augenbraun, L. Anderegg, A. P. Sedlack, and J. M.
Doyle, Sisyphus laser cooling of a polyatomic molecule, Phys. Rev. Lett. 118, 173201 (2017).
34 T. A. Isaev and R. Berger, Polyatomic candidates for cooling of molecules with lasers from
simple theoretical concepts, Phys. Rev. Lett. 116, 063006 (2016).
35 I. Kozyryev, L. Baum, K. Matsuda, and J. M. Doyle, Proposal for laser cooling of complex
polyatomic molecules, ChemPhysChem 17, 3641 (2016).
36 M. V. Ivanov, S. Gulania, and A. I. Krylov, Two cycling centers in one molecule: Communica-
![Page 35: From orbitals to observables and backiopenshell.usc.edu/pubs/pdf/jcp_perspective.pdfQuantum mechanics explains chemical and physical phenomena by using abstract math-ematical constructs,](https://reader034.vdocuments.us/reader034/viewer/2022042406/5f1fa754210ddc366d060e82/html5/thumbnails/35.jpg)
35
tion by through-bond interactions and entanglement of the unpaired electrons, J. Phys. Chem.
Lett. 11, 1297 (2020), in press.
37 M. V. Ivanov, T.-C. Jagau, G.-Z. Zhu, E. R. Hudson, and A. I. Krylov, In search of molecular
ions for optical cycling: A difficult road, Phys. Chem. Chem. Phys. (2020), in press, doi:
10.1039/D0CP02921A.
38 M. V. Ivanov, F. H. Bangerter, P. Wojcik, and A.I. Krylov, Towards ultracold organic chemistry:
Prospects of laser cooling large organic molecules, J. Phys. Chem. Lett. (2020), submitted.
39 N. Turro, Modern Molecular Photochemistry. Benjamin/Cummings Pub. Co., 1978.
40 J. Michl and V. Bonacic-Koutecky, Electronic Aspects of Organic Photochemistry. Wiley, New
York, 1990.
41 In the same sense and with the same caveats as discussed in the case of Dyson orbitals — it can,
in principle, be reconstructed from sufficiently detailed experimental information, i.e., angular
resolved absorption cross sections for molecules fixed in space.
42 S. Tretiak and S. Mukamel, Density matrix analysis and simulation of electronic excitations in
conjugated and aggregated molecules, Chem. Rev. 102, 3171 (2002).
43 S. Tretiak, C. Middleton, V. Chernyak, and S. Mukamel, Exciton Hamiltonian for the bacte-
riochlorophyll system in the LH2 antenna complex of purple bacteria, J. Phys. Chem. B 104,
4519 (2000).
44 M. Head-Gordon, A. M. Grana, D. Maurice, and C. A. White, Analysis of electronic transitions
as the difference of electron attachment and detachment densities, J. Phys. Chem. 99, 14261
(1995).
45 P. R. Surjan, Natural orbitals in CIS and singular-value decomposition, Chem. Phys. Lett.
439, 393 (2007).
46 The state density matrix is Hermitian in the exact theory and can become slightly non-Hermitian
within approximate treatments, such as coupled-cluster and equation-of-motion coupled-cluster
theory. However, since only the Hermitian part of the state density matrix contributes to state
properties, such as dipole moments, one can work with symmetrized state density matrices for
the purpose of orbital analyses and calculations of molecular properties.
47 F. Plasser, M. Wormit, and A. Dreuw, New tools for the systematic analysis and visualization
of electronic excitations. I. Formalism, J. Chem. Phys. 141, 024106 (2014).
48 Promotion numbers computed as the sum of positive (or negative) occupations of the density
![Page 36: From orbitals to observables and backiopenshell.usc.edu/pubs/pdf/jcp_perspective.pdfQuantum mechanics explains chemical and physical phenomena by using abstract math-ematical constructs,](https://reader034.vdocuments.us/reader034/viewer/2022042406/5f1fa754210ddc366d060e82/html5/thumbnails/36.jpg)
36
difference matrix ∆ provide a more general metric. For CIS wave functions, both quantities
yield the same answer. A large doubly excited character of a transition would manifest itself in
a reduced value of Ω, whereas density-difference eigenvalues would deliver a more direct answer
(i.e., ∼2) (see examples in Ref. 53).
49 A. V. Luzanov, A. A. Sukhorukov, and V. E. Umanskii, Application of transition density matrix
for analysis of excited states, Theor. Exp. Chem. 10, 354 (1976), Russian original: Teor. Eksp.
Khim., 10, 456 (1974).
50 A. V. Luzanov and V. F. Pedash, Interpretation of excited states using charge-transfer number,
Theor. Exp. Chem. 15, 338 (1979).
51 R. L. Martin, Natural transition orbitals, J. Phys. Chem. A 118, 4775 (2003).
52 I. Mayer, Using singular value decomposition for a compact presentation and improved inter-
pretation of the CIS wave functions, Chem. Phys. Lett. 437, 284 (2007).
53 F. Plasser, S. A. Bappler, M. Wormit, and A. Dreuw, New tools for the systematic analysis and
visualization of electronic excitations. II. Applications, J. Chem. Phys. 141, 024107 (2014).
54 S. A. Bappler, F. Plasser, M. Wormit, and A. Dreuw, Exciton analysis of many-body wave
functions: Bridging the gap between the quasiparticle and molecular orbital pictures, Phys.
Rev. A 90, 052521 (2014).
55 S. Mewes, F. Plasser, A. I. Krylov, and A. Dreuw, Benchmarking excited-state calculations
using exciton properties, J. Chem. Theory Comput. 14, 710 (2018).
56 P. Kimber and F. Plasser, Toward an understanding of electronic excitation energies beyond
the molecular orbital picture, Phys. Chem. Chem. Phys. 22, 6058 (2020).
57 The analysis of excited states in terms of transition densities and density differences was pio-
neered by Luzanov and co-workers49,50 in the seventies. The term ”Natural Transition Orbitals”
was coined by Martin51, who independently arrived at this concept in 2003, following Amos and
Hall’s work91. He illustrated NTOs by using the CIS ansatz, overlooking the fact that in this
case NTOs are formally and numerically equivalent to the natural orbitals of density difference,
whose squares yield attachment-detachment densities, which were described by Head-Gordon
and co-workers in 199544. NTOs (in the context of CIS) were rediscovered yet again in 2007 by
Mayer52. Responding to Mayer’s paper, Surjan has shown45 that CIS NTOs are equivanlent to
the NOs, again overlooking all prior work on the suject44,49–51. Russian-speaking readers can
enjoy a humorous account of the history of NTOs and SVD in quantum chemistry in the essay
![Page 37: From orbitals to observables and backiopenshell.usc.edu/pubs/pdf/jcp_perspective.pdfQuantum mechanics explains chemical and physical phenomena by using abstract math-ematical constructs,](https://reader034.vdocuments.us/reader034/viewer/2022042406/5f1fa754210ddc366d060e82/html5/thumbnails/37.jpg)
37
by Luzanov92, presented as an illustration of the (extended) Arnold’s principle: ”If a notion
bears a personal name, then this name is not the name of the discoverer”.
58 F. Plasser, Entanglement entropy of electronic excitations, J. Chem. Phys. 144, 194107 (2016).
59 M. Goppert-Mayer, Uber elementarakte mit zwei quantensprungen, Ann. d. Phys. 401, 273
(1931).
60 J. Olsen and P. Jørgensen, Linear and non-linear response functions for an exact state and for
MCSCF state, J. Chem. Phys. 82, 3235 (1985).
61 K. D. Nanda and A. I. Krylov, Two-photon absorption cross sections within equation-of-
motion coupled-cluster formalism using resolution-of-the-identity and Cholesky decomposition
representations: Theory, implementation, and benchmarks, J. Chem. Phys. 142, 064118 (2015).
62 M.J. Paterson, O. Christiansen, F. Pawlowski, P. Jørgensen, C. Hattig, T. Helgaker, and
P. Salek, Benchmarking two-photon absorption with CC3 quadratic response theory, and com-
parison with density-functional response theory, J. Chem. Phys. 124, 054332 (2006).
63 S. Knippenberg, D. R. Rehn, M. Wormit, J. H. Starcke, I. L. Rusakova, A. B. Trofimov, and
A. Dreuw, Calculations of nonlinear response properties using the intermediate state represen-
tation and the algebraic-diagrammatic construction polarization propagator approach: Two-
photon absorption spectra, J. Chem. Phys. 136, 064107 (2012).
64 M. Kallay and J. Gauss, Calculation of frequency-dependent polarizabilities using general
coupled-cluster models, J. Molec. Struct. (THEOCHEM) 768, 71 (2006).
65 K. D. Nanda and A. I. Krylov, A simple molecular orbital picture of RIXS distilled from
many-body damped response theory, J. Chem. Phys. 152, 244118 (2020).
66 M. A. El-Sayed, Spin–orbit coupling and the radiationless processes in nitrogen heterocycles,
J. Chem. Phys. 38, 2834 (1963).
67 M. A. El-Sayed, Triplet state: Its radiative and non-radiative properties, Acc. Chem. Res. 1, 8
(1968).
68 P. Pokhilko and A. I. Krylov, Quantitative El-Sayed rules for many-body wavefunctions from
spinless transition density matrices, J. Phys. Chem. Lett. 10, 4857 (2019).
69 P. A. Malmqvist, B. O. Roos, and B. Schimmelphennig, The restricted active space (RAS) state
interaction approach with spin-orbit coupling, Chem. Phys. Lett. 357, 230 (2002).
70 P. Pokhilko, E. Epifanovsky, and A. I. Krylov, General framework for calculating spin–orbit
couplings using spinless one-particle density matrices: theory and application to the equation-
![Page 38: From orbitals to observables and backiopenshell.usc.edu/pubs/pdf/jcp_perspective.pdfQuantum mechanics explains chemical and physical phenomena by using abstract math-ematical constructs,](https://reader034.vdocuments.us/reader034/viewer/2022042406/5f1fa754210ddc366d060e82/html5/thumbnails/38.jpg)
38
of-motion coupled-cluster wave functions, J. Chem. Phys. 151, 034106 (2019).
71 D. E. Freedman, W. H. Harman, T. D. Harris, G. J. Long, C. J. Chang, and J. R. Long, Slow
magnetic relaxation in a high-spin iron(II) complex, J. Am. Chem. Soc. 132, 1224 (2010).
72 W. P. Reinhardt, Complex coordinates in the theory of atomic and molecular structure and
dynamics, Annu. Rev. Phys. Chem. 33, 223 (1982).
73 N. Moiseyev, Non-Hermitian quantum mechanics. Cambridge University Press, 2011.
74 S. Klaiman and I. Gilary, On resonance: A first glance into the behavior of unstable states,
Adv. Quantum Chem. 63, 1 (2012).
75 T.-C. Jagau, K. B. Bravaya, and A. I. Krylov, Extending quantum chemistry of bound states
to electronic resonances, Annu. Rev. Phys. Chem. 68, 525 (2017).
76 Z. Benda and T.-C. Jagau, Communication: Analytic gradients for the complex absorbing
potential equation-of-motion coupled-cluster method, J. Chem. Phys. 146, 031101 (2017).
77 Z. Benda, K. Rickmeyer, and T.-C. Jagau, Structure optimization of temporary anions, J.
Chem. Theory Comput. 14, 3468 (2018).
78 Z. Benda and T.-C. Jagau, Understanding processes following resonant electron attachment:
Minimum-energy crossing points between anionic and neutral potential energy surfaces, J.
Chem. Theory Comput. 14, 4216 (2018).
79 S. Benda and T.-C. Jagau, Locating exceptional points on multidimensional complex-valued
potential energy surfaces, J. Phys. Chem. Lett. 9, 69786984 (2018).
80 T.-C. Jagau and A. I. Krylov, Characterizing metastable states beyond energies and lifetimes:
Dyson orbitals and transition dipole moments, J. Chem. Phys. 144, 054113 (2016).
81 W. Skomorowski and A. I. Krylov, Real and imaginary excitons: Making sense of resonance
wavefunctions by using reduced state and transition density matrices, J. Phys. Chem. Lett. 9,
4101 (2018).
82 H. Feshbach, A unified theory of nuclear reactions. 2, Ann. Phys. (N.Y.) 19, 287 (1962).
83 W. Skomorowski, S. Gulania, and A. I. Krylov, Bound and continuum-embedded states of
cyanopolyyne anions, Phys. Chem. Chem. Phys. 20, 4805 (2018).
84 F. Schulz, M. Ijas, R. Drost, S. K. Hamalainen, A. Harju, A. P. Seitsonen, and P. Liljeroth,
Many-body transitions in a single molecule visualized by scanning tunnelling microscopy, Nat.
Phys. 11, 229 (2015).
85 J. Itatani, J. Levesque, D. Zeidler, H. Niikura, H. Pepin, J.C. Kieffer, P.B. Corkum, and D.M.
![Page 39: From orbitals to observables and backiopenshell.usc.edu/pubs/pdf/jcp_perspective.pdfQuantum mechanics explains chemical and physical phenomena by using abstract math-ematical constructs,](https://reader034.vdocuments.us/reader034/viewer/2022042406/5f1fa754210ddc366d060e82/html5/thumbnails/39.jpg)
39
Villenueve, Tomographic imaging of molecular orbitals, Nature 432, 867 (2004).
86 R. Santra, Imaging molecular orbitals using photoionization, Chem. Phys. 329, 357 (2006).
87 P. Puschnig, S. Berkebile, A. J. Fleming, G. Koller, K. Emtsev, T. Seyller, J. D. Riley,
C. Ambrosch-Draxl, F. P. Netzer, and M. G. Ramsey, Reconstruction of molecular orbital
densities from photoemission data, Science 326, 702 (2009).
88 S. Wei, D. Luftner, T. Ules, E. M. Reinisch, H. Kaser, A. Gottwald, M. Richter, S. Soubatch,
G. Koller, M. G. Ramsey, F. S. Tautz, and P. Puschnig, Exploring three-dimensional orbital
imaging with energy-dependent photoemission tomography, Nat. Comm. 6, 8287 (2017).
89 M. Graus, C. Metzger, M. Grimm, P. Nigge, V. Feyer, A. Scholl, and F. Reinert, Three-
dimensional tomographic imaging of molecular orbitals by photoelectron momentum microscopy,
Eur. Phys. J. B 92, 80 (2019), https://doi.org/10.1140/epjb/e2019-100015-x.
90 https://qpd.chem.msu.ru/memorial/; assessed on July 25, 2020.
91 A. T. Amos and G. G. Hall, Single determinant wave functions, Phys. Rev. C A 263, 483
(1961).
92 A. V. Luzanov, About V. Arnolds principle and its application to quantum chemistry and
related fields, Kharkov University Bulletin; N. 1026 Chemical Series 21 (2012).