developments in ultrafast x-ray spectroscopy of metal …1g symmetry, and the 4p has t 1u....

14
Developments in ultrafast x-ray spectroscopy of metal complexes Jeroen Ubink March 2016 Abstract An overview of the field of ultrafast spectroscopy of metal complexes is given in this research paper. First, a general treatment of the electronic structure of metal complexes is given, followed by a discussion of x-ray spectroscopy methods and ultrafast spectroscopy in general. Lastly, recent developments in the understanding of the spin crossover mechanism in iron(II) complexes are discussed. Contents 1 Introduction 1 2 Electronic structures and spectra of metal complexes 2 2.1 Crystal-field theory .......................................... 2 2.2 Ligand-field theory .......................................... 3 2.3 Electronic structure of Fe(II) complexes .............................. 5 3 X-ray spectroscopy methods 6 4 Ultrafast x-ray spectroscopy 8 5 Recent developments on spin crossover in iron(II) compounds 9 6 Conclusion 13 1 Introduction Metal complexes are of considerable scientific interest and are studied, for instance, for use in light-harvesting or as dye-sensitisers in solar cells[4, 5, 6], or for their function in biological systems (such as in photosys- tem II[16], or myoglobin[17]). The electronic transitions that determine the interesting properties of metal complexes occur at such a short time scale, that the only way to track their behaviour in time is through ultrafast spectroscopy. In this research paper, an overview will be given of recent developments in ultrafast x-ray spectroscopy of metal complexes. First, basic background information on the electronic structure of metal complexes is treated. Afterwards, an overview of various x-ray spectroscopy techniques, as well as the basic principles of ultrafast spectroscopy, are given. Lastly, recent developments in the study of spin crossover in iron(II) complexes are discussed. 1

Upload: others

Post on 25-Jul-2020

2 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Developments in ultrafast x-ray spectroscopy of metal …1g symmetry, and the 4p has t 1u. Furthermore, two of the 3d orbitals have e g symmetry, and the remaining three 3d orbitals

Developments in ultrafast x-ray spectroscopy of metal complexes

Jeroen Ubink

March 2016

Abstract

An overview of the field of ultrafast spectroscopy of metal complexes is given in this research paper. First,a general treatment of the electronic structure of metal complexes is given, followed by a discussion ofx-ray spectroscopy methods and ultrafast spectroscopy in general. Lastly, recent developments in theunderstanding of the spin crossover mechanism in iron(II) complexes are discussed.

Contents

1 Introduction 1

2 Electronic structures and spectra of metal complexes 22.1 Crystal-field theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22.2 Ligand-field theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32.3 Electronic structure of Fe(II) complexes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

3 X-ray spectroscopy methods 6

4 Ultrafast x-ray spectroscopy 8

5 Recent developments on spin crossover in iron(II) compounds 9

6 Conclusion 13

1 Introduction

Metal complexes are of considerable scientific interest and are studied, for instance, for use in light-harvesting

or as dye-sensitisers in solar cells[4, 5, 6], or for their function in biological systems (such as in photosys-

tem II[16], or myoglobin[17]). The electronic transitions that determine the interesting properties of metal

complexes occur at such a short time scale, that the only way to track their behaviour in time is through

ultrafast spectroscopy. In this research paper, an overview will be given of recent developments in ultrafast

x-ray spectroscopy of metal complexes. First, basic background information on the electronic structure of

metal complexes is treated. Afterwards, an overview of various x-ray spectroscopy techniques, as well as

the basic principles of ultrafast spectroscopy, are given. Lastly, recent developments in the study of spin

crossover in iron(II) complexes are discussed.

1

Page 2: Developments in ultrafast x-ray spectroscopy of metal …1g symmetry, and the 4p has t 1u. Furthermore, two of the 3d orbitals have e g symmetry, and the remaining three 3d orbitals

2 Electronic structures and spectra of metal complexes

Before treating modern developments in ultrafast x-ray spectroscopy of metal complexes, it is helpful to first

give an overview of the basics of metal complex electronic structure. Much of the following discussion on

the electronic structure of metal complexes was taken from [1].

First, the electronic structure of metal complexes will be treated. To understand the electronic structure of

complexes, two models can be used. These models differ in the way they approach metal-ligand interactions.

The first of these models is crystal-field theory. This theory is formally only applicable to ions in crystals,

but it nonetheless gives the basic details of the electronic structure of metal complexes[1]. The other model

is ligand-field theory, which is an application of molecular orbital theory. Ligand-field theory describes the

electronic structure of metal complexes more accurately and in more detail than crystal-field theory does.

Therefore, the predictions for the electronic structure of complexes given by crystal-field theory will be

treated first.

2.1 Crystal-field theory

Crystal-field theory is based on the principle that the ligands near the metal core can be approximated as

negative point charges. The ligands are therefore substituted for an effective field; the so called ligand field1.

These point-charge ligands are bound to the positively charged metal ion core by attractive Coulombic in-

teractions. However, the interactions between the ligands and the metal centre’s electrons are repulsive.

Because of the ligand-electron repulsion, the space surrounding the metal is no longer energetically isotropic.

The orbital configuration of the metal centre therefore loses symmetry. The result is a lifting of the de-

generacies of the metal core’s energy levels. The exact way in which these levels will split depends on the

geometry of the ligands. For an octahedral configuration, the metal’s five d-electron levels will split into a

triply-degenerate t2g orbital, and a doubly-degenerate eg state. The energy spacing between these levels is

called the ligand-field splitting parameter, ∆O, where O denotes that the splitting is due to an octahedral

ligand field[1]. It should be noted that the initial energy level that is split by the octahedral ligand field

is not equal to the energy level of the d-orbitals of a free metal ion. Instead, the ligand-field splitting is

defined with respect to the energy level of the metal ion in a hypothetical spherical ligand field, which has the

same total amount of charge as the ligand field. The energy level of the metal’s d-orbitals in this spherical

field is called the barycentre of the d-orbitals. Another thing to note is that the ligand field splitting is not

symmetrical around the barycentre. For instance, for octahedral ligand field splitting, the three t2g orbitals

are lowered by an amount of 25∆O with respect to the barycentre, and the two eg levels are raised by 2

5∆O

(see figure 1).

The magnitude of ∆O is determined by the identity of the ligands and the metal core. A ligand that

results in a large ligand-field splitting parameter is called a strong-field ligand, and similarly a ligand that

causes a low amount of energy splitting is called a weak-field ligand. The strength of the ligands significantly

influences the spin characteristics of the complex’s electronic structure. The reason for this is the way in

which the split electron levels are filled. The d-electrons of the metal are divided over the two split orbitals

in a way that minimises the total energy of the electron configuration. Using the aufbau principle, the lowest

split orbital will be filled first by electrons with aligned spins. Once all of the low-energy levels are singly

occupied, the next electron can either be placed with an opposing spin in one the low-energy levels, or it

1In the original application of crystal-field theory for modelling ions in crystals, the effective field substitutes the rest of thecrystal and is then called the crystal field.

2

Page 3: Developments in ultrafast x-ray spectroscopy of metal …1g symmetry, and the 4p has t 1u. Furthermore, two of the 3d orbitals have e g symmetry, and the remaining three 3d orbitals

Figure 1: Figure A shows the ligand-field splitting of the d-orbitals of a complex’s metal centre due to anoctahedral field. Figure B and C show the electron configurations in a weak field and a strong field complex,respectively. Image adapted from [2].

can be placed in one of the high-energy levels. Which of the two levels the electron occupies depends on the

magnitude of the ligand-field splitting parameter. If ∆O is larger than the electron pairing energy (i.e., the

energy needed to pair two electrons -with opposing spins- in the same orbital), the next electron will occupy

one of the lower energy levels, whereas if ∆O is smaller than the pairing energy, the electron will occupy

one of the high-energy levels with its spin aligned with the spins of the lower-energy electrons. The result

of this is that low-field ligands (which give a small ∆O) will yield a high-spin electron configuration, while

strong-field ligands give a low-spin configuration (see figure 1).

2.2 Ligand-field theory

While crystal-field theory can explain the existence of ∆O, it cannot predict a value for it. Furthermore, since

ligands are treated only as point charges in crystal-field theory, the theory cannot describe the spectroscopic

features of the ligands themselves. To better describe the electronic structure of metal complexes, it is

necessary to use a different theory: ligand-field theory. This theory is based on molecular orbital theory,

and gives a more accurate representation of the ligands by actually considering the ligands’ orbitals, instead

of treating the ligands as point charges. Following [1], a short description of the main results of ligand-field

theory will be given.

In ligand-field theory, the orbitals of the complex are determined by considering symmetry-adapted linear

combinations (SALCs) of the metal centre’s orbitals with the orbitals of the ligands. There are two possible

ways of bonding: σ-bonding and π-bonding. The exact way in which these orbitals combine depends on

the symmetry of the metal complex. As an example, the SALCs for octahedral complexes will be discussed

3

Page 4: Developments in ultrafast x-ray spectroscopy of metal …1g symmetry, and the 4p has t 1u. Furthermore, two of the 3d orbitals have e g symmetry, and the remaining three 3d orbitals

in this section. First, σ-bonding will be treated. In an octahedrally symmetric complex, the metal core is

surrounded by six ligands; four in the horizontal plane, and two axial ligands. In transition metals, the 3d,

4s, and 4p orbitals make up the frontier orbitals of the system. The 4s orbital has a1g symmetry, and the 4p

has t1u. Furthermore, two of the 3d orbitals have eg symmetry, and the remaining three 3d orbitals have t2g

symmetry. The possible symmetries for the σ-symmetry ligand orbitals are a1g, t1u and eg. Therefore, the

complex’s orbitals will consist of SALCs of the a1g, t1u and eg symmetry orbitals of both metal and ligands,

while the metal’s t2g orbitals do not form SALCs.

The other type of metal-ligand bonding is π-bonding. This type of bonding is due to SALCs of metal and

ligand orbitals with π-symmetry. In octahedral complexes, the metal’s t2g d-orbitals have π-symmetry, and

can therefore form π-bonds with suitable ligands. The interaction between the ligand π-symmetry orbital

and the metal’s t2g orbital results in a splitting of the π and t2g orbital into a bonding and antibonding level.

To see what the effect is on the electronic structure of the complex, it is again helpful to make use of the

aufbau principle. For ligands with fully filled π-symmetry orbitals (known as π-donor ligands), the bonding

level is filled up by the ligands’ electrons. The metal’s t2g electrons then must occupy the antibonding π-bond

level. The result of this, is a decrease in ∆O. For ligands with empty π-symmetry orbitals, the metal’s t2g

electrons can all occupy the bonding level, and the result is an increase in ∆O (see figure 2).

Figure 2: Figure illustrating the influence that π-donor (left) and π-acceptor (right) ligands have on ∆O.Image adapted from [3], such that the diagram illustrates the examples given by [1].

4

Page 5: Developments in ultrafast x-ray spectroscopy of metal …1g symmetry, and the 4p has t 1u. Furthermore, two of the 3d orbitals have e g symmetry, and the remaining three 3d orbitals

2.3 Electronic structure of Fe(II) complexes

Figure 3: Graph depicting a general energy diagramof an octahedral Fe(II) complex. For the electronicstructure of this diagram, the ligands bind to the ironcore through Fe-N bonds. The horizontal axis depictsthe Fe-N bond length elongation. The energy diagramdemonstrates the fact that the Fe-N bond is elongatedupon excitation from the ground state to certain ex-cited state levels. Image adapted from [7].

Now that a general overview of the basics of metal

complex electronic structure has been given in the

previous subsections, it is useful to consider the elec-

tronic structure of one type of metal complex in

more detail. In this section, the electronic struc-

ture of iron(II) complexes will be treated. The dis-

cussion of the electronic structure of Fe(II) in this

section will serve as an illustration of the basic con-

cepts described earlier, and it will provide necessary

background information for the discussion on recent

developments in ultrafast x-ray spectroscopy in sec-

tion 5.

The electronic structure of iron(II) is determined

by the six valence electrons it has in its 3d shell. As

described in section 2.1, the presence of an octahe-

dral ligand- (or crystal-) field causes the d-orbitals

of the metal core of a complex to split into a triply-

degenerate t2g orbital, and a doubly-degenerate eg

orbital. In the ground state configuration of iron(II)

in an octahedral complex (1A1), the six 3d-electrons

all occupy the t2g orbital[7]. Because all electron

spins are paired in this configuration, the ground

state of the metal core is a low-spin (LS) state (see

figure 4). An energy diagram of a general Fe(II)

complex is given in figure 3. The lowest excited

state of an Fe(II) core in an octahedral field is the

quintet 5T2 state[7]. In this state, two of the ground

state’s t2g electrons have been excited to the eg state, resulting in four unpaired electron spins, making this

state a high-spin (HS) state (see figure 4). Other excited electronic states present in diagram are the singlet

and triplet metal-centred 1,3T2 states, and the singlet and triplet metal to ligand charge transfer (MLCT)

states. An MLCT transition is a type of charge-transfer transition, in which an electron is either trans-

ferred from a metal-centred orbital to a ligand-centred orbital (MLCT), or vice versa (ligand to metal charge

transfer (LMCT))[1]. Charge-transfer transitions in d6-metal complexes are triggered by light, which makes

these complexes an interesting material for use in light-harvesting or as dye-sensitisers in solar cells[4, 5, 6].

Moreover, the energy of charge-transfer transitions can be tuned by adjusting or substituting ligands[5].

The electronic structure diagram from figure 3 depicts the energies of various electronic states of an Fe(II)

complex with Fe-N metal-ligand bonds, as a function of Fe-N bond elongation. From this figure, one can see

that the Fe-N bond length of the MLCT state is equal to the ground state bond length. In contrast, the LS

and HS states do have bond elongations; of respectively 0.1 and 0.2 A.

Lastly, Fe(II) complexes exhibit the phenomenon of spin crossover (SCO)[7]. In SCO processes, electrons

are transferred from a low-spin state to a high-spin state or vice versa, and thus the electrons are subject

5

Page 6: Developments in ultrafast x-ray spectroscopy of metal …1g symmetry, and the 4p has t 1u. Furthermore, two of the 3d orbitals have e g symmetry, and the remaining three 3d orbitals

to a spin flip[7]. In Fe(II) complexes, the transition from the LS ground state to the HS 5T2 state is an

SCO process. Depending on the type of polypyridine ligand, SCO can be induced in Fe(II) complexes by,

for instance, temperature, pressure, or light[7]. Recent developments in studies of SCO in iron(II) complexes

are discussed in detail in section 5.

3 X-ray spectroscopy methods

Figure 4: Energy diagram showing how theligand-field split d-orbitals of the Fe(II) coreare occupied for the case of the HS 5T2 ex-cited state and the LS 1A1 ground state.Image adapted from [7].

In this section, the different methods of x-ray spectroscopy will

be treated. First, resonant inelastic x-ray scattering (RIXS)

will be discussed in detail. Since the underlying principles

are the same for all spectroscopy methods, it suffices to treat

only one method in detail. Thus, after having given a detailed

overview of RIXS, the other, absorption-based, methods are

discussed more briefly.

For probing electronic structural information with x-rays,

various techniques exist. These techniques are all very similar

to one another, and differ only in the details of which electronic

transitions they probe. Following the review from [8], the prin-

ciples of RIXS will be discussed, after which the discussion is

extended to the other probing techniques. In RIXS, the sample

being studied is irradiated with x-rays. An x-ray photon ar-

riving on the sample excites electrons from their ground state

level to an unoccupied higher level. In order for this transi-

tion to be possible, the energy of the photon has to equal the

energy of the transition. In other words, the photon needs to

be resonant with the transition. In this excited state, the sys-

tem has a hole in one of its inner energy levels, which is highly

energetically unstable. Consequently, an electron from one of

the higher occupied energy levels will decay to the excited electron’s ground state, thereby emitting a new

x-ray photon. From the energy of the emitted photon, one can study the electronic transitions present in

the sample. The electronic transitions that can be probed are called absorption edges. Transitions from the

1s state are known as K-edges, transitions from n = 2 levels as L edges, and so on.

There are two different RIXS mechanism that result in the re-emission of a photon. Which of the two

mechanisms is responsible for the emission of the outgoing photon depends on the energy of the incoming

x-ray photon, as well as on the strength of the transitions between the core states and the conduction-band

states of the sample[8] The first and simplest of the two mechanism is direct RIXS. In direct RIXS, the core

electron is excited to an unoccupied state in the valence band of the sample. The resulting core hole is then

filled by an electron from the occupied valence band, producing a photon of an energy lower than that of

the excitation photon. This decay results in a hole being left in the valence band. The result of this is that

there is now an electron-hole excitation present in the valence band of the material. This excitation is a

quasiparticle that can propagate through the material. The momentum of the electron-hole excitation is,

according to conservation of momentum, equal to the difference between the momentum of the excitation

x-ray photon and the emitted lower energy photon.

6

Page 7: Developments in ultrafast x-ray spectroscopy of metal …1g symmetry, and the 4p has t 1u. Furthermore, two of the 3d orbitals have e g symmetry, and the remaining three 3d orbitals

The second RIXS mechanism is called indirect RIXS. For the indirect mechanism to occur, the transition

between the core states and the valence band states needs to be weakly allowed, to prevent the system from

following the direct RIXS procedure. Furthermore, the incoming x-ray photon needs to have a high enough

energy to excite a core electron to a state several eV above the Fermi level of the material[8]. If these two

criteria are met, the indirect RIXS process can occur. This process takes place in the following way: first,

an incoming x-ray photon excites a core electron to an energy level several eV above the Fermi level. After

the excitation, the resulting core hole has a Coulombic interaction with electrons in the valence band of

the material. This Coulombic interaction can excite valence electrons to unoccupied valence states, thus

creating a valence band excitation. In the final step, the electron excited by the incoming photon decays

to its original postion to fill up the core hole, thereby emitting a photon. Due the Coulombic interaction

between the core hole and the valence excitation in the intermediate step of inelastic RIXS, the core hole

has lost some energy (i.e. has moved ‘up’ in the energy diagram). The annihilation of this hole by the

excited electron will therefore not emit a photon with an energy equal to that of the excitation photon. The

difference in energy between the incoming and outgoing photon is equal to the energy of the valence band

excitation created during the RIXS process. In systems where both direct and inelastic RIXS processes are

allowed, x-ray scattering will be largely due to the direct mechanism, with the indirect process occurring

only as higher order contributions[8]. Lastly, it should be noted that different authors use the terms RIXS,

RXES (resonant x-ray emission spectroscopy), and XES (x-ray emission spectroscopy) in different ways, and

sometimes interchangeably, to refer to the same kinds of processes[8, 9, 10, 11]. To avoid confusion, the term

‘XES’ will be used in the remainder of this paper to refer to both RIXS and RXES processes.

The second kind of x-ray spectroscopy is based on the absorption of x-ray photons. These absorption-

based techniques share many similarities with XES, and will be discussed in the remainder of this section.

The term x-ray absorption spectroscopy (XAS), is usually used to describe two kinds of absorption

spectroscopy[11]: x-ray absorption near edge structure (XANES), and extended x-ray absorption fine struc-

ture (EXAFS). XANES spectra are taken in a range of -20 to +100 eV with respect to an absorption edge[11],

and therefore XANES spectra record pre- and post-edge features. For EXAFS spectra, the incoming x-ray

photons have sufficient energy to excite core electrons to the continuum. The EXAFS signal coming from

one scattering atom is given by the difference between the photoelectron wave emitted from the scattering

atom and the backscattered waves coming from the neighbouring atoms. Because of this dependence on

neighbouring atoms, it is possible to extract data on interatomic spacings and orientations from EXAFS

signals[11]. XAS spectra can be measured in two ways: in transmission mode, in which the absence of the

absorbed photons is measured, or in fluorescence mode, in which the photoemission from the refilling of the

core hole is measured.

XAS spectra measured in fluorescence mode are very similar to XES spectra, but they provide subtly

different information about electronic strucutre[11]. In XES, the photoemission is due to core-hole refilling

from electrons in occupied energy levels higher than the core hole, but lower than the excited electron.

Therefore, the XES signal will not be at the wavelength of the excitation photon, but instead the signal

will be at wavelengths corresponding to transitions between the occupied energy levels and the core level.

Therefore, in XES, the intensity of the signals is proportional to the density of states in the occupied higher

energy levels[11]. In contrast, for XAS, the photoemission is due to the filling up of the core-hole by the

excited electron itself. What then determines the strength of a XAS signal is whether it is possible to excite

an electron to an energy level ~ω higher in energy. In other words, the signals in XAS are proportional to

the densities of states of the unoccupied energy levels[11].

7

Page 8: Developments in ultrafast x-ray spectroscopy of metal …1g symmetry, and the 4p has t 1u. Furthermore, two of the 3d orbitals have e g symmetry, and the remaining three 3d orbitals

4 Ultrafast x-ray spectroscopy

In this section, a short discussion of ultrafast x-ray techniques will be given. First, the general principles of

ultrafast x-ray spectroscopy will be discussed. Afterwards, a brief overview of two sources of x-ray pulses -

synchrotrons and free-electron laser - is given.

Ultrafast spectroscopy studies are usually performed with a so-called pump-probe method. In this tech-

nique, a sample is first exposed to a pump pulse, which excites the process that one is interested in (e.g.

electronic transitions, or phonon modes). Next, after a certain time delay, a second pulse - the probe pulse -

is applied to the sample. The probe pulse then interacts with the excitation caused by the pump pulse. After

interacting with the sample, the scattered probe pulse is studied with a detector, which provides information

on the properties of the excitation. By varying the time delay between the pump and probe pulses, and

measuring the probe pulse for every value of the time delay, one gains multiple ‘snapshots’ of the excited

system. These snapshots provide information on how the excited state develops over time.

In order to study excited state processes such as electronic transitions and charge transfer, the time

resolution of the pump-probe system needs to be on the same order of magnitude as the time scale on which

the process of interest occurs. Depending on the type of process, the time scale varies between 10−6 and

10−15 seconds[11]. The time scale of pump probe systems is determined by a convolution of two factors: the

pulse duration of the longest lasting of the two pulses, and the instrument response function of the optical

equipment used to measure the probe pulse[11]. However, the instrument response function only plays a role

when the pulse durations are on the same order of magnitude as the instrument’s response time.

Ultrafast x-ray pulses (i.e. in the order of picoseconds to femtoseconds) can be obtained from two sources:

synchrotrons, and free electron lasers. In the remainder of this section, the principles of the operation of

these two sources will be discussed in brief.

First, x-rays from synchrotrons will be treated. Synchrotrons are particle accelerators for which the

particles’ trajectory is kept fixed during acceleration by increasing the strength of the applied magnetic

fields proportional to the velocity of the particles[12]. Due to the electrons’ circular motion, they emit

electromagnetic radiation. By moving the electrons in a train of multiple ‘bunches’ through the synchrotron’s

storage ring, one can obtain short x-ray pulses at a fixed frequency. For details on synchrotrons as x-ray

sources, the reader is referred to reference [13]. Lastly, reference [14] gives an overview of how femtosecond

x-ray pulses can be obtained from a synchrotron source by means of a ‘slicing’ technique.

X-ray pulses can also be obtained from free electron lasers (FELs)[15]. In FELs, electron are accelerated

through a magnetic undulator. A sinusoidally alternating magnetic field is applied between the undulator,

which causes the electrons to oscillate. Due to mutual interactions, the electrons will end up oscillating in

phase (hence the name ‘free electron laser’). These in-phase electrons can be used to generate x-ray pulses.

For more details on FELs as x-ray pulse sources, see reference [15].

8

Page 9: Developments in ultrafast x-ray spectroscopy of metal …1g symmetry, and the 4p has t 1u. Furthermore, two of the 3d orbitals have e g symmetry, and the remaining three 3d orbitals

5 Recent developments on spin crossover in iron(II) compounds

Ultrafast x-ray spectroscopy is used to study many different kinds of processes and systems. It is used, for

instance, to study biological systems, such as photosystem II[16], or myoglobin[17]. For references to more

recent studies involving ultrafast x-ray spectroscopy, the reader is referred to the review by Chen et al.[11].

Figure 5: Energy diagram showing the SCOmechanism in Fe(II) complexes proposed byDecurtins et al.. Note that Decurtins et al.do not use the term ‘MLCT’ in labellingthe energy levels (compare this image withfigure 3). Figure taken from [18].

Due to the breadth of the kinds of topics studies with ultra-

fast x-ray spectroscopy, a complete overview of recent work on

all these different topics is beyond the scope of this research pa-

per. Instead, an in-depth description of one topic in particular

- spin crossover in iron(II) complexes - will be given.

As described in section 2.3, spin crossover transitions can

occur in iron(II) complexes. In SCO transitions, electrons go

from a low spin state to a high spin state. SCO compounds

are being studied for their potential application in magnetic

data storage and nanoscale data processing, and because SCO

processes are important to the binding of ligands in heme

proteins[22]. In order to more consciously design compounds

for these applications of SCO compounds, it is necessary to

gain detailed knowledge about the energy levels and geometri-

cal configuration for the states relevant to the SCO process. In

Fe(II) compounds, the intermediate energy states that play a

role in the SCO process have lifetimes in the order of picosec-

onds or less. Ultrafast techniques are therefore needed to study

spin crossover in Fe(II) compounds.

It was reported[19] in 1980 by Sutin et al.[20] and in 1983

by Netzl et al.[21] that, after photoexcitation of the ground

state to the 1MLCT state, the HS state in iron(II) complexes

becomes occupied with a quantum yield of unity. In 1985,

Decurtins et al. observed an electronic transition from the LS1A1 ground state to the HS 5T2 excited state (see figure 5)

in multiple iron(II) complexes[18]. They found that below a

temperature of 50K, the electrons could be ‘trapped’ into the

HS state and kept there upon continuous irradiation of the sample. They coined this phenomenon light-

induced excited state spin trapping (LIESST). The transition mechanism from the LS to the HS state,

proposed by Decurtins et al., consisted of an excitation from the 1A1 ground state to a singlet excited

MLCT state (1T2), followed by an intersystem crossing from the singlet MLCT state to the lowest triplet

excited state facilitated by spin-orbit coupling, and lastly a decay from the triplet state to the quintet HS

state[18].

At first glance, it is odd that the SCO process in Fe(II) complexes should occur with unity quantum yield

when the process could involve two intersystem crossings[19]. One would expect the intermediate singlet and

triplet states in the SCO process described earlier to have strong electronic coupling to the ground state,

and therefore that the electron would more probably decay to the ground state instead of to the 5T2 state.

However, this coupling apparently does not prevent the electron from decaying to the HS state.

9

Page 10: Developments in ultrafast x-ray spectroscopy of metal …1g symmetry, and the 4p has t 1u. Furthermore, two of the 3d orbitals have e g symmetry, and the remaining three 3d orbitals

Since the paper by Decurtins et al., much research has been conducted on SCO in iron(II) complexes[7].

Ultrafast x-ray spectroscopy is a vital tool for examining the excited state structure and dynamics of metal

complexes. The reason why x-ray methods, rather than optical methods, are more useful for these kinds of

studies is that optical spectroscopy has two distinct disadvantages2 for studying metal complex electronic

structure[11]. Firstly, visible range signals from metal cores in complexes are weaker than the signals from

optical transitions in aromatic ligands, making it difficult to distinguish between the two types of signals.

Secondly, the excited states of some metal cores lie outside the visible light range (i.e. they are optically

dark), and therefore cannot be probed with optical light. With the use of x-ray spectroscopy, one does not

have to deal with these disadvantages.

Ultrafast x-ray absorption spectroscopy (XAS) can yield valuable information on the fine electronic

structure of metal centres in complexes. In XANES spectra, features just below the Fermi level (so-called pre-

edge features) give information on oxidation state, occupancy of valence orbitals, charge transfer, and other

electronic details of the metal complex[7]. Furthermore, XANES provides information on bond distances and

angles, and coordination numbers as well[7]. XAS can also study features above the Fermi level: EXAFS

spectra can give insights into bond distances and coordination numbers of the nearest neighbour atoms

surrounding the x-ray absorbing atom[7].

Figure 6: A ball-and-stick model of[Fe(bipy)3]2+. Image taken from [25].

There has been a considerable amount of research on spin

crossover in iron(II)-tris-bipyridine ([Fe(bipy)3]2+) [22, 23, 24,

25]. The goal of all of these studies was to determine the exact

SCO pathway in this complex. There was especially a debate

on whether or not the 1,3T2 functioned as intermediate states

in the LS → HS transition[23]. In the remainder of this sec-

tion, a series of recent developments will be given on SCO in

[Fe(bipy)3]2+).

In 2007, Chergui et al. performed picosecond x-ray absorp-

tion spectroscopy measurements on SCO in [Fe(bipy)3]2+[22].

Using both EXAS and XANES, they found that the Fe-N bond

increases by about 0.2 A going from the LS ground state to the

HS excited state. They were not able to monitor the struc-

tural changes in the compound for the entire SCO transition,

because the intermediate states in the SCO process decayed on

the sub-picosecond timescale, which their picosecond XAS could not probe[22].

The same research group performed another study of [Fe(bipy)3]2+ in 2009, but this time with a better

time-resolution made possible by the installation of a beam-slicing technique (see section 4) at the Swiss

Light Source (SLS, PSI-Villigen)[23]. The new equipment allowed them to have a < 250 fs time resolution.

In their experiments, the group used a combination of an optical excitation pump pulse and an x-ray probe

pulse. The optical pulse excited electrons from the ground state of the iron(II) complex to the 1MLCT state.

Chergui et al. observed a strong absorption feature in the spectrum of the HS state that was not present

in the LS state’s spectrum. The signal at this feature, which they called the ‘B-feature’, is proportional to

the Fe-N bond length[23]. Therefore, by tracking the transient behaviour of the B-feature, the group was

able to monitor the structural evolution in the SCO process. Based on the results from their experiments

2These two disadvantages are adapted from[11].

10

Page 11: Developments in ultrafast x-ray spectroscopy of metal …1g symmetry, and the 4p has t 1u. Furthermore, two of the 3d orbitals have e g symmetry, and the remaining three 3d orbitals

and theoretical fits to their data, Chergui et al. concluded that the 1,3T2 states are not occupied during the

SCO process. Their results thus suggest that the SCO process in [Fe(bipy)3]2+ occurs as 1A1 → 1MLCT

→ 3MLCT → 5T2, skipping the 1,3T states. As demonstrated by earlier publications [22, 18], SCO was

assumed by some to involve the 1,3T states. However, this assumption was still the subject of debate[23, 25],

and remained so after the 2009 paper by Chergui et al.[25].

Figure 7: Figure 7(A) depicts the XANESspectra of the LS and HS state in[Fe(bipy)3]2+ (for a representation of themolecular structure of the complex, see fig-ure 6). The HS spectrum was obtained aftera 50 ps time delay with respect to the laserexcitation. It should be noted that the HSspectrum was not measured directly, butwas obtained from combining the LS spec-trum and the transient spectrum in (B) (fordetails on how the HS spectrum was deter-mined, the reader is referred to references[22] and [23]). Figure 7(B) shows the tran-sient difference XANES spectrum (see ref-erence [22]) between the sample excited bythe laser pump and the unexcited sample.The red dots show the spectrum recordedat 50 ps after excitation, and the stars showthe spectrum 300 fs after excitation. The B-feature shows up as a distinct peak in thetransient spectrum. Image taken from [23].

In 2013, Cammarata et al. studied [Fe(bipy)3]2+ as

well, but using x-ray pulses from a free-electron laser (from

the Linac Coherent Light Source (LCLS)) instead of from a

synchrotron[24]. The XFEL produced x-ray pulses of less than

100 fs, which, combined with an instrument response time of

150 fs, is an improvement over the <250 fs time resolution

of Chergui et al. in 2009[23]. The improved time resolution

allowed Cammarata et al. to more accurately measure the life-

time of the LS → HS transition. This lifetime was found to be

about 160 fs[24]. However, the authors were not able to con-

clude anything about the role of the 1,3T2 states in the SCO

process.

The group of Gaffney et al. published a paper on

[Fe(bipy)3]2+ in 2014, in which they studied the complex by ex-

amining the transient Kβ fluoresence spectrum of the iron(II)

core[25]. Kβ fluorescence was used to study the complex be-

cause that technique can resolve the spin state of the compound

(see figure 8), and thereby more easily identify the intermedi-

ate states in the SCO process. However, Kβ fluorescence is

not sensitive to the electronic details of the ligand and the lo-

cal symmetry of the complex, and therefore cannot be used to

distinguish between the singlet and triplet MLCT states[25].

Gaffney et al. recorded spectra of [Fe(bipy)3]2+, and fitted two

theoretical models through the spectra they obtained. In the

first model, the 1,3MLCT state decayed directly to the HS 5T2

state. In the second model, a 3T intermediate state was in-

cluded. Based on these fits, Gaffney et al. could conclude with

95% certainty that the SCO process includes a 3T intermediate

state[25]. They furthermore explain that a transition involving

a 3T intermediate would, based on theory, be more reasonable

than the direct 1,3MLCT → 5T2 transition. With a transition

including the 3T state, all transitions involve only a single electron, and the different states can all be cou-

pled by a first-order spin-orbit operator[25]. In contrast, the direct transition would involve a transition of

two different electrons on two different centres at the same time, which cannot be mediated by a first-order

spin-orbit operator[25]. Gaffney et al. conclude their paper by stating that the Kβ spectroscopy techniques

they used could not directly observe the presence of the 3T state. Thus, in order to conclusively settle the

debate on the SCO pathway in iron(II) complexes, a spectroscopy techniques that is highly sensitive to the3T state needs to be used.

11

Page 12: Developments in ultrafast x-ray spectroscopy of metal …1g symmetry, and the 4p has t 1u. Furthermore, two of the 3d orbitals have e g symmetry, and the remaining three 3d orbitals

The results by Chergui et al. from 2009[23] and the results by Gaffney et al. in 2014[25] thus contradict

one another. Based on the results of the former, the SCO pathway in [Fe(bipy)3]2+ is expected to occur

without any intermediate states, while the results of the latter indicate with high precision that SCO does

involve a 3T intermediate state. The reason why Chergui et al. concluded that the 3T state was absent, is

that a theoretical fit through their data would suggest a <60 fs lifetime for the 3T state, which they consider

to be unrealistically short[23]. The group of Gaffney et al. also based their conclusion on a fit of a theoretical

model through measured spectra. Based on their fit, the 3T should be present in the SCO pathway. However,

both groups only indirectly determined the presence or absence of the 3T state. My personal view on the

subject is that the results by Gaffney et al. are more likely to be correct. Their analysis of the data allowed

them to state with 95% certainty that the 3T state is present. Chergui et al., in contrast, cannot quantify the

certainty of their conclusion. Their argument for the absence of the intermediate state is that the inclusion

of it would mean that the SCO process would consist of multiple ultrafast inter-system crossing steps, which

Chergui et al. consider unlikely. However, the absence of the 3T state in the SCO process would mean

that electrons are somehow able to transfer with unity quantum yield from an MLCT state to the HS state,

while these states do not have a strong coupling between them (see the arguments given by Gaffney et al.

discussed above). In my opinion, this would make it difficult to have the high quantum yield reported by

Decurtins et al.[18]. I therefore personally side with Gaffney et al. in the debate over the SCO pathway in

iron(II) complexes, but until the presence or absence of the 3T state is directly observed, I do not expect

the debate to be over.

Figure 8: Figure a shows the different shape of Kβ fluorescence signals based on the spin state of thecomplex. For each data set, a different complex with a different spin ground state was used. Figure b showsthe difference spectrum (see reference [22]) between the singlet and doublet, triplet, and quintet spectra froma. The clear distinction between the various difference spectra in b demonstrates that Kβ fluorescence couldbe used to differentiate between the difference spectra of the [Fe(bipy)3]2+ ground state and the 1,3MLCT,3T2 and 5T2 states. Image adapted from [25].

12

Page 13: Developments in ultrafast x-ray spectroscopy of metal …1g symmetry, and the 4p has t 1u. Furthermore, two of the 3d orbitals have e g symmetry, and the remaining three 3d orbitals

6 Conclusion

Ultrafast x-ray spectroscopy is a useful technique for the study of metal complexes. Absorption based

techniques, such XANES or EXAFS, as well as elastic or inelastic scattering techniques give an insight into

how a metal complex’s electronic and geometrical structure change during an electronic transition. The

ongoing debate on spin crossover in iron(II) complexes shows that the field of ultrafast x-ray spectroscopy

of metal complexes still holds many problem that need to be solved. The development of newer, even faster

x-ray spectroscopy techniques in the future will undoubtedly provide new insights and open up new fields of

study.

References

[1] P. Atkins, T. Overton, J. Rourke, M. Weller, and F. Armstrong, Shriver & Atkins’ Inorganic Chemistry,

2010, Oxford University Press, 473-504.

[2] Robert J. Lancashire, Crystal Field Stabilization Energy, retrieved from http://chemwiki.

ucdavis.edu/Core/Inorganic_Chemistry/Crystal_Field_Theory/Introduction_to_Crystal_

Field_Theory/Crystal_Field_Stabilization_Energy in March 2016.

[3] Beverly J. Volicer and Steven F. Tello, d-Metal Complexes, retrieved from http://faculty.uml.edu/

ndeluca/84.334/topics/topic6.htm in March 2016.

[4] S. Ardo and G. J. Meyer, Chem.Soc.Rev., 2009, 38, 115–164.

[5] A. S. Polo, M. K. Itokazu, N. Y. M. Iha, Coordination Chemistry Reviews 248 (2004) 1343–1361.

[6] K. Kalyanasundaram and M. Gratzel, Coordination Chemistry Reviews 177 (1998) 347–414.

[7] A. Cannizzo, C. J. Milne, C. Consani, W. Gawelda, Ch. Bressler, F. van Mourik and M. Chergui,

Coordination Chemistry Reviews 254 (2010) 2677–2686.

[8] L. J. P. Ament, M. van Veenendaal, T. P. Devereaux, J. P. Hill and J. van den Brink, Rev. Mod. Phys.,

2011 Vol. 83, 705-767.

[9] M. Rovezzi and P. Glatzel, Semicond. Sci. Technol. 29 (2014) 023002.

[10] A. Kotani, K. O. Kvashnina, S. M. Butorin and P. Glatzel, Eur. Phys. J. B, (2012) 85: 257

[11] L. X. Chen, X. Zhang and M. L. Shelby, Chem. Sci., 2014, 5, 4136.

[12] C. Zhang and S.X. Fang, Handbook of Accelerator Physics and Engineering, edited by A. W. Chao, K.

H. Mess, M. Tigner and F. Zimmermann, 2013, World Scientific, 58-59.

[13] K. Holldack, R. Ovsyannikov, P. Kuske, R. Muller, A. Schalicke, M. Scheer, M. Gorgoi, D. Kuhn, T.

Leitner, S. Svensson, N. Martensson and A. Fohlisch, 2014, Nature Communications 5, Article number:

4010.

[14] R. W. Schoenlein, S. Chattopadhyay, H. H. W. Chong, T. E. Glover, P. A. Heimann, C. V. Shank, A.

A. Zholents and M. S. Zolotorev, Science New Series, Vol. 287, No. 5461 (Mar. 24, 2000), pp. 2237-2240.

13

Page 14: Developments in ultrafast x-ray spectroscopy of metal …1g symmetry, and the 4p has t 1u. Furthermore, two of the 3d orbitals have e g symmetry, and the remaining three 3d orbitals

[15] C. Pellegrini and J. Stohr, X-Ray Free Electron Lasers: Principles, Properties and Applications, re-

trieved from https://www-ssrl.slac.stanford.edu/stohr/xfels.pdf in March 2016.

[16] H. Dau and M. Haumann, Photosynth. Res. (2007) 92:327–343.

[17] F. A. Lima, C. J. Milne, D. C. V. Amarasinghe, M. H. Rittmann-Frank, R. M. van der Veen, M.

Reinhard, V.-T. Pham, S. Karlsson, S. L. Johnson, D. Grolimund, C. Borca, T. Huthwelker, M. Janousch,

F. van Mourik, R. Abela and M. Chergui, Rev. Sci. Instrum., 2011, 82, 063111.

[18] S. Decurtins; P. Gutlich, K. M. Hasselbach, A. Hauser and H. Spiering, Inorg. Chem. 1985, 24, 2174-

2178.

[19] M. Chergui, Spin-Crossover Materials : Properties and Applications, edited by M. A. Halcrow, Somerset,

NJ, USA: John Wiley & Sons, 2013, 405-424.

[20] C. Creutz, M. Chou, T. L. Netzel, M. Okumura and N. Sutin, J. Am. Chem. Soc., 1980, 102 (4), pp

1309–1319.

[21] M. A. Bergkamp, C. Chang and T. L.Netzel, J. Phys. Chem. 1003, 87, 4441-4446.

[22] W. Gawelda, V.-T. Pham, M. Benfatto, Y. Zaushitsyn, M. Kaiser, D. Grolimund, S. L. Johnson, R.

Abela, A. Hauser, C. Bressler and M. Chergui, Phys. Rev. Lett., 2007, 98, 057401.

[23] C. Bressler, C. Milne, V. T. Pham, A. ElNahhas, R. M. v. d. Veen, W. Gawelda, S. Johnson, P. Beaud,

D. Grolimund, M. Kaiser, C. N. Borca, G. Ingold, R. Abela and M. Chergui, Science, 2009, 323, 489–492.

[24] H. T. Lemke, C. Bressler, L. X. Chen, D. M. Fritz, K. J. Gaffney, A. Galler, W. Gawelda, K. Haldrup,

R. W. Hartsock, H. Ihee, J. Kim, K. H. Kim, J. H. Lee, M. M. Nielsen, A. B. Stickrath, W. Zhang, D.

Zhu and M. Cammarata, J. Phys. Chem. A, 2013, 117, 735–740.

[25] W. Zhang, R. Alonso-Mori, U. Bergmann, C. Bressler, M. Chollet, A. Galler, W. Gawelda, R. G. Hadt,

R. W. Hartsock, T. Kroll, K. S. Kjaer, K. Kubicek, H. T. Lemke, H. W. Liang, D. A. Meyer, M. M.

Nielsen, C. Purser, J. S. Robinson, E. I. Solomon, Z. Sun, D. Sokaras, T. B. van Driel, G. Vanko, T.-C.

Weng, D. Zhu and K. J. Gaffney, Nature, 2014, 509, 345–348.

14