150
Photoinduced Spin Crossover in Single Crystal [Fe II (bpy) 3 ](PF 6 ) 2 by Ryan Lucas Field A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Graduate Department of Physics University of Toronto c Copyright 2016 by Ryan Lucas Field

[FeII(bpy) ](PF - University of Toronto T-Space · 2017. 3. 27. · aqueous [FeII(bpy) 3] 2+ are also present in the crystal case. However, oscillations observed on this time scale

  • Upload
    others

  • View
    8

  • Download
    0

Embed Size (px)

Citation preview

  • Photoinduced Spin Crossover in Single Crystal[FeII(bpy)3](PF6)2

    by

    Ryan Lucas Field

    A thesis submitted in conformity with the requirementsfor the degree of Doctor of Philosophy

    Graduate Department of PhysicsUniversity of Toronto

    c© Copyright 2016 by Ryan Lucas Field

  • Abstract

    Photoinduced Spin Crossover in Single Crystal [FeII(bpy)3](PF6)2

    Ryan Lucas Field

    Doctor of Philosophy

    Graduate Department of Physics

    University of Toronto

    2016

    Spin crossover (SCO) is a phenomenon in which the spin state of a molecular system

    is changed by photoexcitation or a change in temperature/pressure. Research in SCO

    has been driven by potential applications in data storage and solar energy generation.

    SCO also has biological relevance, as it is involved in the function of certain metallopro-

    teins. [FeII(bpy)3]2+ is an SCO system that undergoes a low-spin (LS) to high-spin (HS)

    transition following photoexcitation. In aqueous solutions, this process is extremely fast,

    with the HS state being populated within ∼100 fs of the initial excitation. While the

    dynamics of this system have been extensively studied in solution, comparatively little

    research has been done in the solid phase, which would be a requirement for potential

    applications.

    Herein, the ultrafast SCO dynamics in single crystal [FeII(bpy)3](PF6)2 are character-

    ized using transient absorption (TA) spectroscopy to elucidate the effects of the crystal

    environment on the SCO process. This environment is of particular interest for probes

    sensitive to the nuclear motions involved in the SCO transition such as time-resolved

    electron diffraction. To this end, a TA spectrometer capable of collecting high-quality

    data over a broad spectral range in both liquid and solid samples was developed.

    The results show that the ultrafast dynamics following excitation are very similar

    those in aqueous [FeII(bpy)3]2+, while dynamics on longer time scales are perturbed in

    the crystal environment. The spectral signatures associated with subpicosecond SCO in

    ii

  • aqueous [FeII(bpy)3]2+ are also present in the crystal case. However, oscillations observed

    on this time scale suggest more complicated nuclear motions than those observed in aque-

    ous samples. At longer times after excitation, the HS state decays back to the LS ground

    state almost an order of magnitude faster than in the aqueous case because of chemical

    pressure exerted by the lattice. The effects of a long-range acoustic phonon also manifest

    themselves as periodic modulations of the absorption peaks. It is concluded that SCO in

    single crystal [FeII(bpy)3](PF6)2 is a local molecular process similar to that in aqueous

    [FeII(bpy)3]2+, with effects specific to the crystal environment primarily manifesting long

    after the relaxation to the HS state is complete.

    iii

  • Acknowledgements

    I must first and foremost thank my advisor R. J. Dwayne Miller. Dwayne’s enthusiasm

    in the pursuit of scientific endeavors has been inspirational in my development as a

    scientist. He has a reassuring optimism even when things go awry and an unwavering

    belief in his students’ capabilities. Throughout my long and sometimes tumultuous PhD

    work, he never seemed to doubt my ability to carry this work through to completion.

    That has made all the difference.

    All the students past and present who have passed through the Miller group during

    my years here should be acknowledged for their support and camaraderie. I would like

    to extend particular thanks to a number of individuals who played an important role

    in bringing this work to fruition. My close friend and collaborator Nelson Liu has been

    constant source of support, both theoretical and emotional, throughout my PhD work.

    I’m not sure how I would have gotten through it without him. I would also like to thank

    former graduate students Alexei Halpin and Philip Johnson. During their time in the

    group, they were always available to provide guidance when needed. The excellent quality

    of their work inspired me to stay focused in my own efforts. Etienne Pelletier should

    also be acknowledged for providing great technical advice during the development of my

    lab. I thank Olivier Paré-Labrosse for constructive discussions that helped guide the

    interpretation of my results. Yifeng Jiang has been a close collaborator during the final

    few years of this work, and contributed the ultrafast electron diffraction data that appears

    in the final chapter. Patrick Rui also collaborated on experiments done in Hamburg and

    contributed to the development of the apparatus used in this work in the early days.

    Finally, Alessandra Picchiotti was a great friend and host during my visits to Germany,

    and made my time there very enjoyable.

    The Miller group’s postdocs and research associates have been a great help throughout

    the years. I was particularly lucky to have worked directly with Valentyn Prokhorenko

    during the early years of my PhD work. It was from him that I first learned the particulars

    iv

  • of working in an ultrafast laser lab. His relentless pursuit of exceptionally clean data

    provided the benchmark that I tried to attain in developing my own setup. Even after

    he moved to the German side of the Miller group in Hamburg, he continued to provide

    invaluable advice throughout the development of the experimental apparatus used in

    this work. Research associate Gustavo Moriena has been consistently understanding

    and helpful, and has always made sure that the larger aspects of our shared lab space

    were kept under control. Postdocs Amy Stevens, Samansa Maneshi, and Henrike Müller-

    Werkmeister have been very supportive in recent years, both scientifically and personally.

    Prior to the arrival of the current team of postdocs, Arash Zarrine-Afsar, Francis Talbot,

    Germán Sciaini, and Ryan Cooney helped keep me motivated and on track.

    Thanks must be extended to contributors outside of the Miller group. Working from

    the European XFEL, Wojciech Gawelda provided the samples used for the experiments

    in this thesis, and provided valuable insights in interpreting my data. Former Miller

    group postdoc and frequent collaborator at the University of Toronto Cheng Lu did the

    local sample preparation necessary to make the experiments presented herein possible. I

    would also like to thank my advisory committee, University of Toronto professors John

    Wei and Dvira Segal, as well as McGill University professor Patanjali Kambhampati who

    acted as the external examiner at my final PhD defense.

    Finally, my deepest thanks go to my family and friends. Jessica Gahunia has been a

    constant source of love and support over the last few years. Her perseverance and work

    ethic in her own studies has influenced my own habits and kept me focused throughout

    the writing process. I would also like to thank my sister and brother-in-law Dawn and

    Geoff Dinnes and my nephews Nathan and Taylor who I look forward to watching grow

    in the coming years. Most importantly, my parents John and Heather Field have always

    encouraged me in my academic endeavors, and have always been there to keep me positive

    and grounded even in the toughest times. I could not have done it without them.

    v

  • Contents

    1 Spin Crossover in Condensed Matter 1

    1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

    1.1.1 Ultrafast Measurement . . . . . . . . . . . . . . . . . . . . . . . . 1

    1.1.2 Spin Crossover . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

    1.1.3 [FeII(bpy)3]2+ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

    1.2 Magnetic Characterization of Spin States . . . . . . . . . . . . . . . . . . 7

    1.2.1 Magnetic Susceptibility . . . . . . . . . . . . . . . . . . . . . . . . 7

    1.2.2 Mössbauer Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . 9

    1.3 UV-Vis spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

    1.3.1 UV-Vis Absorption Spectrum of [FeII(bpy)3]2+ . . . . . . . . . . . 12

    1.3.2 Time-Resolved UV-Vis Studies of [FeII(bpy)3]2+ . . . . . . . . . . 13

    1.4 Ultrafast Time-Resolved X-Ray Studies . . . . . . . . . . . . . . . . . . . 17

    1.5 Time-Resolved Studies of Related Spin Crossover Crystals . . . . . . . . 23

    1.5.1 Cooperative Lattice Effects . . . . . . . . . . . . . . . . . . . . . 25

    1.6 Overview and Thesis Outline . . . . . . . . . . . . . . . . . . . . . . . . 27

    2 Spin Crossover Theory 29

    2.1 Electronic Structure of [FeII(bpy)3]2+ . . . . . . . . . . . . . . . . . . . . 29

    2.2 Thermal Spin Crossover . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

    2.3 Photoinduced Spin Transitions . . . . . . . . . . . . . . . . . . . . . . . . 35

    vi

  • 2.3.1 Internal Conversion . . . . . . . . . . . . . . . . . . . . . . . . . . 36

    2.3.2 Intersystem Crossing . . . . . . . . . . . . . . . . . . . . . . . . . 38

    2.3.3 Application to [FeII(bpy)3]2+ . . . . . . . . . . . . . . . . . . . . . 40

    3 Experimental Setup 43

    3.1 Femtosecond Laser System . . . . . . . . . . . . . . . . . . . . . . . . . . 43

    3.2 Second Harmonic Generation . . . . . . . . . . . . . . . . . . . . . . . . 45

    3.3 Pulse Shaping and Characterization . . . . . . . . . . . . . . . . . . . . . 47

    3.3.1 4F Acousto-Optic Pulse Shaper . . . . . . . . . . . . . . . . . . . 48

    3.3.2 Limitations of Acousto-Optic Pulse Shapers . . . . . . . . . . . . 52

    3.3.3 Frequency-Resolved Optical Gating . . . . . . . . . . . . . . . . . 54

    3.4 Broadband Supercontinuum Generation . . . . . . . . . . . . . . . . . . . 58

    3.5 Transient Absorption Spectrometer . . . . . . . . . . . . . . . . . . . . . 62

    3.5.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

    3.5.2 Data Collection . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

    3.5.3 Beam Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . 66

    3.5.4 Group Velocity Dispersion Correction and Estimation of Instru-

    ment Response Function Duration . . . . . . . . . . . . . . . . . . 68

    4 Ultrafast Dynamics of Single Crystal [FeII(bpy)3](PF6)2 71

    4.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71

    4.2 Experimental Details . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72

    4.2.1 Sample Preparation . . . . . . . . . . . . . . . . . . . . . . . . . . 72

    4.2.2 Measurement Conditions . . . . . . . . . . . . . . . . . . . . . . . 73

    4.3 Linear Absorption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75

    4.4 Transient Absorption Data . . . . . . . . . . . . . . . . . . . . . . . . . . 76

    4.4.1 Linearity of the TA Signal . . . . . . . . . . . . . . . . . . . . . . 76

    4.4.2 TA of Aqueous [FeII(bpy)3]2+ . . . . . . . . . . . . . . . . . . . . 78

    vii

  • 4.4.3 TA of Single Crystal [FeII(bpy)3](PF6)2 . . . . . . . . . . . . . . . 79

    4.5 Method of Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82

    4.6 Results of Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84

    4.6.1 Aqueous [FeII(bpy)3]2+ . . . . . . . . . . . . . . . . . . . . . . . . 84

    4.6.2 Single Crystal [FeII(bpy)3](PF6)2 . . . . . . . . . . . . . . . . . . 88

    4.7 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99

    5 Conclusions and Future Work 102

    5.1 Development of the Experiment . . . . . . . . . . . . . . . . . . . . . . . 102

    5.2 Summary of [FeII(bpy)3]2+ Results . . . . . . . . . . . . . . . . . . . . . . 103

    5.2.1 Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105

    5.3 Future Work on [FeII(bpy)3](PF6)2 . . . . . . . . . . . . . . . . . . . . . 106

    5.3.1 Ultrafast Electron Diffraction . . . . . . . . . . . . . . . . . . . . 106

    5.3.2 Time-Resolved Magneto-Optic Kerr Effect Spectroscopy . . . . . 108

    5.4 Future Applications of the Experimental Setup: Coherent Control . . . . 109

    5.4.1 Candidate System: ZW-NAIP . . . . . . . . . . . . . . . . . . . . 110

    5.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112

    Bibliography 114

    viii

  • List of Figures

    1.1 a) Unit cell of the LS [FeII(bpy)3](PF6)2 crystal. The atoms are coloured

    as follows: orange - iron, blue - nitrogen, black - carbon, purple - phos-

    phorous, green - fluorine. b) Structures of the LS (blue) and HS (red)

    [FeII(bpy)3]2+ ion, showing the elongated Fe–N bonds in the HS state. . . 7

    1.2 The product χMT (where χM is the molar magnetic susceptibility and T

    is temperature) as a function of temperature in [Fe(btz)2(NCS)2] (left),

    [Fe(phen)2(NCS)2] (middle) and [Fe(dppz)2(NCS)2]pyridine (right). Re-

    produced from [14] with permission of The Royal Society of Chemistry. . 8

    1.3 a) Mössbauer spectrum of [Fe(ptz)6](BF4)2 at various temperatures, show-

    ing a thermal spin transition around 136 K. The blue shaded areas are due

    to the LS complex, while the red shaded areas are due to the HS complex.

    Reproduced with permission from [42]. b) Logarithm of decay rates (kHL

    in s-1) as a function of temperature in [Mn:Fe(0.05%)II(bpy)3](PF6). Black

    circles are obtained from time-differential Mössbauer spectroscopy. White

    circles are obtained from optical measurements. Reproduced from [33]

    with permission of Springer. . . . . . . . . . . . . . . . . . . . . . . . . . 9

    ix

  • 1.4 a) Electronic potential energy surfaces of [FeII(bpy)3]2+ as a function of

    Fe–N distance, as described in [24, 25]. b) Absorption spectrum of aque-

    ous [FeII(bpy)3]2+ showing the contribution of various bands at different

    wavelengths. Figures a) and b) reprinted from [17] by permission from

    Macmillan Publishers Ltd.: Nature Chemistry, c© 2015. c) Polarized ab-

    sorption spectrum of [Zn1-xFex(bpy)3](PF6)2 at 293 K using probe polar-

    ization perpendicular to (π) and in the same plane as (σ) the molecular

    trigonal axis of symmetry. Adapted from [13] with permission of Springer. 13

    1.5 Transient absorption of aqueous [FeII(bpy)3]2+ in the UV (a) and visible

    (b) as reported in [17]. Oscillations are clearly visible in both ranges

    after photoexcitation. Reprinted from [17] by permission from Macmillan

    Publishers Ltd.: Nature Chemistry, c© 2015. . . . . . . . . . . . . . . . . 16

    1.6 a) Relaxation rate constants as a function of temperature for the HS state

    of [FeII(bpy)3]2+ doped in various isostructural host lattices, [M(bpy)3](PF6)2

    where M = Co (inverted filled triangle), Zn (filled diamond), Mn (filled tri-

    angle), Cd (filled circle), and in the oxalate network [NaRh(ox)3][Zn(bpy)3]

    (filled square) at ambient pressure, and for the Cd host lattice at 1 kbar

    external pressure (open circle). Inset: Low temperature relaxation rate

    constants as a function of unit cell volume. Reproduced with permission

    from [34], c© 2002, Schweizerische Chemische Gesellschaft. b) Schematic

    diagram showing the mechanism of the different HS→LS [FeII(bpy)3]2+

    relaxation rates through the size of the host lattice’s unit cell (left) and

    the lattice’s effect on the potential energy curves (right). Here, ∆E0HL is

    the energy gap between the minima of the LS and HS state surfaces, and

    ∆QHL in the difference in the position of the minima of the two surfaces

    along the reaction coordinate, Q. Reproduced from [13] with permission

    of Springer. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

    x

  • 1.7 a) Relative change in X-ray absorption at selected energies, along with

    associated fits. The contributions of the MLCT state are shown in orange.

    b) An example fit at E = 7121.5 eV. The top panel shows the contribu-

    tions of the MLCT and HS states, along with the overall signal, with the

    inset showing the change of the signal on a longer time scale. The middle

    panel shows the populations of the MLCT and HS states derived from

    the fit. The bottom panel shows the modeled change in the Fe–N radius,

    noting a 50 fs phase shift of the oscillations (relative to directly excited

    oscillations, shown in light blue) resulting from the convolution with the

    modeled MLCT population. The shaded area shows the ensemble distri-

    bution of Fe–N radii in the vibrationally hot sample, whose width decays

    over time. Reproduced from [22] with permission of Dr. H. T. Lemke. . . 20

    1.8 Cooperative effects in single crystal [Fe(phen)2NCS2] a) HS fraction as a

    function of temperature, showing an abrupt transition at 183 K with a 2 K

    hysteresis loop. Reproduced with permission from [64], c© 2003, published

    by Elsevier Masson SAS. All rights reserved. b) Ultrafast photoswitching

    monitored by transient optical transmission (OT) at 950 nm. c) HS frac-

    tion following photoexcitation. Arrows indicate rises caused by different

    mechanisms. From left to right: ultrafast photoswitching, elastic lattice

    expansion, and thermal switching from laser-deposited heat. d) Acoustic

    phonons generated by elastic strain on the picosecond time scale, moni-

    tored by transient reflection at 950 nm. Adapted from [62] with permission

    of Dr. R. Bertoni. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

    xi

  • 2.1 The five 3d orbitals. The lobes contain 90% of the total electron probabil-

    ity. The colour denotes the phase of the wavefunction. Orange is positive

    and blue is negative. Nodal planes, where the electron probability is zero,

    are also shown. Image from [65] used under Creative Commons by-nc-sa

    3.0 license (http://creativecommons.org/licenses/by-nc-sa/3.0/). . . . . . 31

    2.2 a) Separation of orbital energy levels resulting from ligand field splitting,

    redrawn from [18]. b) Orbital occupancy in the LS and HS states, giving

    and S = 0 (t2g6eg

    0) LS state and a S = 2 (t2g4eg

    2) HS state. . . . . . . . 32

    2.3 a) The HS fraction as a function of temperature for various coupling

    strengths, J2/RTc. Fixed model parameters are used to determine the

    value of ∆G0. In this figure n is used in place of nH and J/kTc is used

    in place of J2/RTc. Reproduced from [66] with permission of Springer. b)

    Model of free energy as a function of the HS fraction at various temper-

    atures when third-order interactions are considered. Tc = 120K in this

    figure. Reprinted from [68], c© 1983, with permission from Elsevier. . . . 34

    2.4 The deactivation mechanism of [FeII(bpy)3]2+ from 1MLCT to 5T2 pro-

    posed in [25]. From [25], used with permission of Wiley-VCH. c© 2013

    Wiley-VCH Verlag 17548 GmbH & Co. KGaA, Weinheim. . . . . . . . . 42

    3.1 Schematic of the AOPS setup. G - Grating, SM - Spherical mirror, FM -

    Folding mirror, AOM - Acousto-optic modulator. The magnitude of the

    AOM Bragg angle, curvature of the spherical mirrors, and visual change

    in colour across the spectrum are exaggerated for clarity. . . . . . . . . . 51

    xii

  • 3.2 a) Schematic of the TG FROG setup. RR - retroreflectors, BS1 - 30R/70T

    fused silica beamsplitter, BS2 - 50R/50T fused silica beamslitter, FS -

    Fused silica windows, OAPM - Off-axis parabolic mirror (f = 50 mm), FSS

    - Fused silica sample, CL - Collimating lens (f =70 mm), AL - Achromatic

    lens (f = 70mm). b) Phase matching diagram and beam geometry at

    OAPM (bottom) and CL (top). . . . . . . . . . . . . . . . . . . . . . . . 56

    3.3 Measured FROG traces (left column) with corresponding retrieved spec-

    tral profiles (middle column) and temporal profiles (right column). Both

    intensity and phase profiles (blue and green lines, respectively) are shown.

    The center column also shows the spectral intensity and phase profiles ex-

    pected from the applied pulse shaping mask (black and red dashed lines

    respectively). a) TL pulse. b) Positively chirped pulse. c) Pulse with

    sinusoidally modulated phase. d) Double pulse. . . . . . . . . . . . . . . 57

    3.4 Schematic of supercontinuum generation setup. HWP - Half-wave plate,

    PBS - Polarizing beam splitting cube, QWP1 - 800 nm quarter-wave plate,

    BBO - β-barium borate crystal for SHG, F1 - 400 nm bandpass filter,

    QWP2 - 400 nm quarter-wave plate, FM - Flippable mirror, L - Lens (f

    = 100 mm), CaF2 - Rotated CaF2 window, OAPM - Off-axis parabolic

    mirror (f = 101.6 mm), AQWP - Achromatic quarter-wave plate (for UV

    or Vis range), F2 - Colour filter (350 nm cutoff or 335-610 nm bandpass

    filter). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

    3.5 Typical spectra of the 400 nm and 800 nm pumped white light continua

    used for the experiments presented. The sharp cutoff of the 400 nm

    pumped WL spectrum at 350 nm is due to the low pass filter used to

    remove the residual 400 nm pump light. . . . . . . . . . . . . . . . . . . 62

    xiii

  • 3.6 Schematic of the TA spectrometer. BS1 - 30R/70T fused silica beam-

    splitter, BS2 - 10R/90T fused silica beamsplitter, P - Polarizer, HWP -

    Half-wave plate, FS - Fused silica window, SM - Spherical mirror (f = 250

    mm), PD - Photodiode, SP - Spectrometer. . . . . . . . . . . . . . . . . . 64

    3.7 Beam profiles of the focused pump, visible probe, and UV probe at the

    sample position measured by knife-edge. The horizontal (blue) and ver-

    tical (red) profiles are shown for each beam, with the circles showing the

    measured data points, and the solid lines showing the fits. The extracted

    1/e2 diameters are shown in the legends. . . . . . . . . . . . . . . . . . . 67

    3.8 Example of time-zero delay correction using measured cross-phase modu-

    lation in deionized water. Positive changes in absorption are shown in red,

    and negative changes are shown in blue. . . . . . . . . . . . . . . . . . . 68

    3.9 Measured cross-phase modulation in deionized water with associated fits.

    Data are shown for the UV (top) and visible (bottom) ranges, with pos-

    itive changes in absorption shown in red, and negative changes shown in

    blue. The rightmost column shows the fitted Gaussian width across all

    wavelengths. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69

    4.1 Measured absorption spectra of aqueous [FeII(bpy)3]2+ and single crystal

    [FeII(bpy)3](PF6)2 samples used in the presented TA experiments. The

    dashed line shows the division between the data measured using the UV

    and visible range supercontinuua. . . . . . . . . . . . . . . . . . . . . . . 75

    4.2 UV TA spectrum of aqueous [FeII(bpy)3]2+ at 5 ps for various pump flu-

    ences (left) along with linear fits at selected wavelengths (right). . . . . . 77

    4.3 UV TA spectrum of single crystal [FeII(bpy)3](PF6)2 at 5 ps for various

    pump fluences (left) along with linear fits at selected wavelengths (right). 78

    4.4 Aqueous TA data in the UV (top) and visible (bottom) ranges, on short

    (-1–4 ps, left) and long (4–850 ps, right) time scales. . . . . . . . . . . . . 79

    xiv

  • 4.5 Single crystal TA data in the UV (top) and visible (bottom) ranges, on

    short (-1–4 ps, left) and long (4–850 ps, right) time scales. . . . . . . . . 80

    4.6 Single crystal TA data in the visible range from 4–850 ps for samples with

    varying optical densities. The measured optical density at 533 nm is shown

    for each of the samples. Each of the data sets shown are normalized for

    better comparison of the apparent oscillatory behavior on this time scale. 81

    4.7 Global fits of the TA data for aqueous [FeII(bpy)3]2+ in the UV (top) and

    visible (bottom) ranges, on short (-1–4 ps, left) and long (4–850 ps, right)

    time scales. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85

    4.8 Decay-associated spectra used in the fits shown in figure 4.7. The DAS

    are shown for the short (left) and long (right) time delay scans in the UV

    (top) and visible (bottom). . . . . . . . . . . . . . . . . . . . . . . . . . . 86

    4.9 Kinetic traces of TA of aqueous [FeII(bpy)3]2+ at selected wavelengths with

    associated fits. Short (left) and long (right) time delay scans are shown. . 87

    4.10 Residuals of the global fits of the aqueous [FeII(bpy)3]2+ TA data in the

    UV (top) and visible (bottom) ranges, on short (-1–4 ps, left) and long

    (4–850 ps, right) time scales. . . . . . . . . . . . . . . . . . . . . . . . . . 88

    4.11 a) Spectrally resolved FT of aqueous [FeII(bpy)3]2+ fit residuals after t =

    300 fs in the UV (top) and visible (bottom) ranges. The FT in the visible

    range is multiplied by 10 for visual clarity. b) Spectrally integrated FT in

    the UV (top) and visible (bottom) ranges. . . . . . . . . . . . . . . . . . 89

    4.12 Global fits of the TA data for single crystal [FeII(bpy)3](PF6)2 in the UV

    (top) and visible (bottom) ranges, on short (-1–4 ps, left) and long (4–850

    ps, right) time scales. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90

    4.13 Decay-associated spectra used in the fits shown in figure 4.12. The DAS

    are shown for the short (left) and long (right) time delay scans in the UV

    (top) and visible (bottom). . . . . . . . . . . . . . . . . . . . . . . . . . . 91

    xv

  • 4.14 Comparison of visible-range long time delay DAS used in the fits of the

    single crystal [FeII(bpy)3](PF6)2 TA data with the GS absorption and its

    second derivative. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92

    4.15 Kinetic traces of TA of single crystal [FeII(bpy)3](PF6)2 at selected wave-

    lengths with associated fits. Short (left) and long (right) time delay scans

    are shown. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92

    4.16 Residuals of the global fits of the single crystal [FeII(bpy)3](PF6)2 TA data

    in the UV (top) and visible (bottom) ranges, on short (-1–4 ps, left) and

    long (4–850 ps, right) time scales. . . . . . . . . . . . . . . . . . . . . . . 94

    4.17 a) Spectrally resolved FT of single crystal [FeII(bpy)3](PF6)2 fit residuals

    after t = 300 fs in the UV (top) and visible (bottom) ranges. The mag-

    nitude of the FTs in the UV and visible ranges are on the same scale. b)

    Spectrally integrated FT in the UV (top) and visible (bottom) ranges. . . 95

    4.18 a) Spectrally resolved FT of single crystal [FeII(bpy)3](PF6)2 fit residuals

    for the visible-range long time delay data shown in figure 4.6. The optical

    density of each sample at 533 nm is inset. b) Corresponding spectrally

    integrated FT for each sample. In each case, the fitted peak corresponds

    to the apparent wavelength-modulated oscillations visible in that data. . 97

    4.19 Experimental phonon periods as a function of measured optical density at

    533 nm in multiple single crystal [FeII(bpy)3](PF6)2 samples. . . . . . . . 98

    xvi

  • 5.1 Preliminary UED data of [FeII(bpy)3](PF6)2 a) Ground state (LS) electron

    diffraction pattern. A section of the data is obscured by a beam blocker

    put in place to block the residual undiffracted electron beam. b) Kinetic

    SVD components of measured changes following excitation with a 400 nm,

    70 fs pulse. The sum of the 2nd and 3rd SVD components is also shown.

    c) Average change in the diffraction pattern corresponding to the sum

    of the 2nd and 3rd SVD components between 2 and 35 ps. The size of

    the circles represents the intensity of the diffraction spot at that position.

    The colour represents the percent change of the diffraction spot with red

    being positive and blue being negative. d) Simulated HS-LS difference

    diffraction pattern. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107

    5.2 a) Ground state absorption spectrum of ZW-NAIP, showing multiple tran-

    sitions as elucidated in [136]. The molecular structure is inset. b) Proposed

    photocycle of ZW-NAIP showing the contributions of various processes to

    the TA spectrum. Adapted from [137] with permission of The Royal So-

    ciety of Chemistry. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110

    5.3 a) Preliminary GVD-corrected TA spectrum of ZW-NAIP in methanol. b)

    Kinetic traces of TA at selected wavelengths. ∼60 cm-1 Oscillations are

    clearly visible in the 434 nm trace. . . . . . . . . . . . . . . . . . . . . . 111

    xvii

  • List of Abbreviations

    AOM Acousto-optic modulator

    AOPS Acousto-optic pulse shaper

    AQWP Achromatic quarter-wave plate

    AWG Arbitrary waveform generator

    BBO β-Barium borate

    CCD Charge coupled device

    CPM Cross-phase modulation

    CW Continuous wave

    DAQ Data acquisition card

    DAS Decay-associated spectrum

    ESA Excited state absorption

    EXAFS Extended X-ray absorption fine structure

    FM Folding mirror

    FROG Frequency-resolved optical gating

    FT Fourier transform

    FWHM Full width at half maximum

    GA Global analysis

    GSB Ground state bleach

    GVD Group velocity dispersion

    HS High-spin

    HWP Half-wave plate

    xviii

  • IC Internal conversion

    IR Infrared

    IRF Instrument response function

    ISC Intersystem crossing

    LBO Lithium triborate

    LCM Liquid crystal modulator

    LIESST Light-induced excited spin state trapping

    LMCT Ligand-to-metal charge-transfer

    LS Low-spin

    MC Metal-centered

    MLCT Metal-to-ligand charge-transfer

    Nd:YLF Neodymium:yttrium lithium fluoride

    Nd:YVO4 Neodymium:yttrium orthovanadate

    OAPM Off-axis parabolic mirror

    OT Optical transmission

    PBS Polarizing beam splitting cube

    PSA Product state absorption

    QWP Quarter-wave plate

    REGEN Regenerative amplifier

    SCO Spin crossover

    SHG Second harmonic generation

    SE Stimulated emission

    SO Spin-orbit

    SQUID Superconducting quantum interference device

    SVD Singular value decomposition

    TA Transient absorption

    TG Transient grating

    xix

  • Ti:Sapph Titanium:sapphire

    TL Transform limited

    TMC Transition metal complex

    TR MOKE Time-resolved magneto-optic Kerr effect spectroscopy

    UED Ultrafast electron diffraction

    UV Ultraviolet

    Vis Visible

    WLG White light generation

    XANES X-ray absorption near edge structure

    XES X-ray emission spectroscopy

    ZW-NAIP Zwitterionic photoswitch based on an N-alkylated indanylidene

    pyrroline Schiff base framework

    xx

  • Chapter 1

    Spin Crossover in Condensed Matter

    1.1 Introduction

    1.1.1 Ultrafast Measurement

    Throughout the history of science, the advancement of technology has been a driver of

    new discoveries. As these advances push the limits of what we are able to observe, we

    gain a deeper and more complete understanding of the world around us.

    To fully understand the dynamics of chemical reactions and other molecular transi-

    tions, it is necessary to resolve their behavior on the relevant time and length scales.

    These scales are, respectively, the femtosecond (10-15 s) time scale and angstrom (10-10

    m) length scale [1]. Typical molecular transitions occur in tens to thousands of femtosec-

    onds, and can involve motions as small as a fraction of an angstrom. In recent decades,

    a number of techniques have become available to directly probe changes in the nuclear

    and electronic properties of molecules on these scales.

    Advances in laser technology have allowed shorter and shorter time scales to be

    probed. Specifically, these improvements are the result of developments allowing very

    short laser pulses to be produced. Currently, laser pulses as short as a few femtosec-

    onds can be generated with relative ease on tabletop setups [2], and recent advancements

    1

  • Chapter 1. Spin Crossover in Condensed Matter 2

    continue to push this limit into the attosecond regime [3]. Shorter pulses improve the

    achievable time resolution by reducing the duration of the interaction between the pulses

    and the material under study.

    Subnanometer spatial resolutions have been achievable through X-ray crystallography

    since as early as 1913 [4], owing to the wavelength of X-rays, which ranges from 0.01 to

    10 nm. Similar resolutions have been achievable with electron microscopes since the

    1950s with recent advancements allowing for spatial resolutions well under an angstrom

    [5, 6]. This resolution is made possible by the effective wavelength (i.e., the de Broglie

    wavelength) of the electrons, which is a function of their kinetic energy and is well below

    an angstrom for kinetic energies as low as 1 keV.

    Time-resolved optical experiments allow various molecular processes to be character-

    ized depending on the wavelength of the probing light, and measurement technique used.

    The most common of these is the “pump-probe” technique, whereby a sample is excited

    by an intense “pump” pulse and subsequently measured using a weaker “probe” pulse

    after a known time delay. Variations of this technique allow different processes to be

    observed. Transitions between electronic states can be followed by monitoring changes in

    absorption of light in the ultraviolet (UV) and visible (Vis) spectral ranges. Fluorescence

    decays can be observed by monitoring gated emitted light from an excited sample. Vibra-

    tional dynamics can be characterized by monitoring changes in the absorption of infrared

    (IR) light. Further information can be obtained from higher-dimensional spectroscopies

    in both the UV-Vis and IR ranges, where sequences involving more pulses can be used

    to directly observe couplings between electronic/vibrational states and to separate the

    contributions of homogeneous and inhomogeneous broadening on the spectrum [7,8].

    In recent years, a number of pump-probe techniques have become available that enable

    measurements with both high temporal and spatial resolution. Ultrafast X-ray diffrac-

    tion, which uses X-ray probe pulses from sources such as synchrotron and free-electron

    laser radiation, allows for changes in the electron density of molecules to be observed,

  • Chapter 1. Spin Crossover in Condensed Matter 3

    which is a proxy for their nuclear structure. Ultrafast electron probe pulses, generated

    by shining femtosecond UV laser pulses on a charged photocathode, can similarly be

    used for time-resolved diffraction experiments [1]. In contrast to X-rays, electrons scat-

    ter primarily from a sample’s atomic nuclei via the Coulomb potential, and thus provide

    a direct probe of nuclear structure. Time-resolved electron diffraction has a number of

    advantages over X-ray diffraction, as it is orders of magnitude less expensive to develop,

    and can be built on a table top. However, additional factors must be taken into account,

    such as the effect of the electrons’ Coulombic repulsion on the achievable pulse duration,

    the intensity or effective brightness of the electron source, and the requirement of a high

    vacuum over the propagation volume of the electrons [1].

    1.1.2 Spin Crossover

    The focus of this thesis is on a molecular transition known as a spin crossover (SCO). This

    refers to a transition whereby the spin multiplicity of a molecule is switched from low-

    spin (LS) to high-spin (HS) or vice-versa by a change in pressure or temperature, or by

    photoexcitation. This phenomenon was first reported in 1931 by Cambi and Szegö, who

    observed the behavior in tris(N,N-dialkyldithiocarbamatoiron(III)) complexes [9]. SCO

    represents an example of molecular multistability, where a molecule has multiple stable

    or metastable electronic states depending on the environmental conditions. As such,

    SCO systems have been studied as potential candidates for a number of applications,

    such as magnetic data storage, solar power generation, optical displays, and molecular

    switches [10]. Various SCO complexes have displayed properties that are conducive

    to such applications, including high efficiency, reversibility, chemical stability, and very

    fast response times. Additionally, SCO has biological relevance, as the active site of

    certain porphyrins, contained in proteins such as hemoglobin and myoglobin, can be

    considered to undergo SCO during the detachment of O2 and the other simple ligands

    they transport [11].

  • Chapter 1. Spin Crossover in Condensed Matter 4

    SCO is commonly observed in various transition metal complexes (TMCs, also called

    coordination complexes). These complexes consist of a metal center called the coordina-

    tion center, surrounded by bound molecules or ions, called ligands. In particular, SCO

    has most commonly been observed in TMCs whose central metal atom is a first-row

    transition metal, although this is not a strict requirement [12].

    SCO is possible because of the energetic structure of the metal atom’s orbitals in

    the presence of the field generated by the ligands. In simple terms, external perturba-

    tions change the electronic configuration of the metal’s d orbitals, which changes the spin

    multiplicity. In general, this change in the electronic configuration is known to be accom-

    panied by structural changes. In particular, the metal-ligand bonds shorten or lengthen

    due to greater bonding or anti-bonding character of the d orbitals occupied [13]. This is

    also known to be associated with a change in the molecular volume and, in the case of

    SCO crystals, unit cell volume [12].

    Thermal spin transitions in LS ground state complexes occur in systems where ther-

    mal energy is sufficient to overcome the energy gap between the spin states. As the

    temperature increases, the thermal energy is converted to successively higher vibrational

    energy levels until a crossing point is reached. Once the gap has been overcome, the

    higher energy spin state is entropically favorable, because of the higher spin multiplicity

    and density of vibrational states [14].

    Direct spin transitions induced by photoexcitation are dipole-forbidden, as the spin

    number is not conserved. Such transitions therefore proceed via relaxation from higher-

    lying excited states. Following photoexcitation, an electron is excited from the metal’s d

    orbitals to the ligand orbitals (or vice-versa, depending on the system [15]), which puts the

    molecule in a so-called metal-to-ligand charge-transfer state, or MLCT (in the opposite

    case, this is called a ligand-to-metal charge-transfer state, LMCT). From there, the spin

    transition occurs through a relaxation process involving a number of intersystem crossing

    (ISC) and possibly internal conversion (IC) steps, which refer to radiationless transitions

  • Chapter 1. Spin Crossover in Condensed Matter 5

    between electronic states having different or the same spin multiplicities, respectively.

    Some of these steps may occur after the electron has returned to the metal’s d orbitals.

    Such states are called “metal-centered” (MC). The specific states in the relaxation process

    depend on the SCO complex under study.

    Following the spin transition, SCO complexes typically relax back to the ground

    state on a longer time scale. At cryogenic temperatures, the photoinduced spin state

    can become “trapped” for long periods of time, as the molecules do not possess enough

    energy to overcome the barrier between the initial and final spin states. This is called

    “light-induced excited spin state trapping” or LIESST. In some cases, the trapped state

    can also be pumped back to the initial state by irradiating at a different wavelength, a

    process called reverse-LIESST [16].

    1.1.3 [FeII(bpy)3]2+

    Some SCO systems do not have accessible thermal transitions, because the energy gap

    between the minima of the LS and HS potential energy surfaces is too large. In these

    cases, spin transitions are only possible via photoexcitation. Studying such systems offers

    unique experimental challenges. Determining the spectral and structural characteristics

    of the LS and HS states is less straightforward than in the opposite case, where these

    properties can be unambiguously determined simply by carrying out experiments at

    temperatures below and above the spin transition temperature.

    One such system is iron(II) tris(2,2’-bipyridine) ([FeII(bpy)3]2+), which is the main

    experimental focus of this thesis. This system undergoes a LS→HS transition upon pho-

    toexcitation by near-UV and visible light [11]. It has been the focus of many studies and

    is sometimes considered the archetype of photoinduced SCO transitions [17]. It serves as

    an excellent model system for the complex interplay between electronic and structural

    degrees of freedom in SCO transitions [18]. As this system lacks a thermal transition,

    its spectral and structural properties must be determined through time-resolved meth-

  • Chapter 1. Spin Crossover in Condensed Matter 6

    ods that can probe the HS state at very short times after it has been populated by

    photoexcitation.

    A noteworthy aspect of the [FeII(bpy)3]2+ complex is the rate at which the HS state

    is formed following photoexcitation. It has been argued that this transition is complete

    within 50 fs [17]. The associated elongation of the central Fe–N bonds is believed to

    occur on a similar time scale [19–22]. This extremely fast transition rate has provoked a

    debate about the relaxation pathway that leads to the final HS state, with inconsistencies

    in the literature regarding precisely which intermediate states are involved [17,23–25].

    Previous studies on this system have investigated its ultrafast dynamics in either

    aqueous samples [17–21, 23, 26–32], [FeII(bpy)3]2+ complexes doped in isostructural host

    crystals [33–37], or in one case, powder crystals of [FeII(bpy)3](PF6)2 [38]. In this thesis,

    SCO is investigated for the first time in single crystals of [FeII(bpy)3](PF6)2 with mea-

    surements on aqueous [FeII(bpy)3]2+ serving as a control experiment for the effects of

    the crystal environment and possible collective effects. The single crystal environment

    is of great interest for measurements that directly probe changes in the molecular struc-

    ture such as time-resolved X-ray and electron diffraction. Studying this phase may also

    provide insights into the behavior of the complex in the solid state environment, which

    would be required for its potential applications. The unit cell of the LS [FeII(bpy)3](PF6)2

    crystal and the structure of the LS and HS states of the [FeII(bpy)3]2+ ion are shown in

    figure 1.1. The structures shown are reconstructed from the results in [39] and [40].

    In this chapter, the literature on SCO will be reviewed, with a focus on [FeII(bpy)3]2+.

    Methods that directly measure the molecular spin state will be discussed, along with

    results identifying the HS state of the [FeII(bpy)3]2+ complex as the metastable state fol-

    lowing photoexcitation. This will be followed a review of results on [FeII(bpy)3]2+ from

    various optical and X-ray spectroscopies, with particular attention on time-resolved spec-

    troscopies. The effects of crystal environments on SCO dynamics will also be discussed,

    along with results on related crystalline SCO systems.

  • Chapter 1. Spin Crossover in Condensed Matter 7

    Figure 1.1: a) Unit cell of the LS [FeII(bpy)3](PF6)2 crystal. The atoms are coloured asfollows: orange - iron, blue - nitrogen, black - carbon, purple - phosphorous, green - fluorine. b)Structures of the LS (blue) and HS (red) [FeII(bpy)3]

    2+ ion, showing the elongated Fe–N bondsin the HS state.

    1.2 Magnetic Characterization of Spin States

    In characterizing the spin states of a molecular complex, direct observations of its mag-

    netic properties are most ideal. Such measurements provide an unambiguous probe, but

    are generally limited in their time resolution. Therefore, such experiments are of little

    utility in characterizing intermediate and transient states in SCO systems. However,

    other methods, such as optical spectroscopies, can be used in conjunction with magnetic

    methods to characterize other properties of the LS and HS states. These can in turn be

    used as indirect probes of the spin states, often having significant advantages such as an

    improved time resolution and signal-to-noise ratio.

    1.2.1 Magnetic Susceptibility

    The most straightforward way to detect changes in a system’s spin state is by direct mea-

    surement of its magnetic susceptibility. In solid samples, superconducting quantum in-

    terference device (SQUID) magnetometers are most commonly used for this purpose [12].

  • Chapter 1. Spin Crossover in Condensed Matter 8

    Figure 1.2: The product χMT (where χM is the molar magnetic susceptibility and T is tem-perature) as a function of temperature in [Fe(btz)2(NCS)2] (left), [Fe(phen)2(NCS)2] (middle)and [Fe(dppz)2(NCS)2]pyridine (right). Reproduced from [14] with permission of The RoyalSociety of Chemistry.

    These magnetometers use superconducting circuits to detect very small magnetic fields,

    as low as a few attotesla [41]. However, they have very low time resolution. For example,

    measurements with the aforementioned sensitivity require several days of accumulation

    time, and as such are not useful in determining transient spin states. Nonetheless, these

    and similar devices are very useful for identifying changes in the magnetic states of SCO

    complexes having thermal transitions, as the different spin states can be populated in-

    definitely, provided the temperature is correctly controlled.

    Figure 1.2 shows examples of the thermal spin transition in three solid SCO com-

    pounds, [Fe(L)2(NCS)2] where L = 2,2’-bithiazoline (btz), 1,10-phenanthroline (phen)

    and dipyrido[3,2-a:2’3’-c]phenazine (dppz), characterized by measurements of the mag-

    netic susceptibility. Using this method, the spin transition temperature can be deduced,

    along with certain characteristics of the transition (i.e., slope, hysteresis, states at in-

    termediate temperatures, etc.). In the solid state, the precise nature of the thermal

    transition is determined by intermolecular cooperativity. This behavior arises from elas-

    tic interactions resulting from the change in volume upon spin transition, leading to

    internal pressure in the solid samples [14]. Gradual thermal transitions (such as in figure

  • Chapter 1. Spin Crossover in Condensed Matter 9

    1.2, left) are observed in samples where there is little cooperativity, whereas abrupt tran-

    sitions, and those with hysteresis (such as in figure 1.2, middle and right, respectively)

    are indicative of highly cooperative systems [12].

    1.2.2 Mössbauer Spectroscopy

    The spin state of SCO compounds can also be directly measured by Mössbauer spec-

    troscopy. This technique uses the recoilless absorption and emission of gamma rays from

    solid crystals, along with the Doppler shift induced by accelerating the absorber/emitter

    through a range of velocities, to measure very small differences in transition energies [42].

    In the case of SCO, this is applied to measure the hyperfine splitting of spin sub-states

    in an external magnetic field. An example of the Mössbauer spectra of a spin crossover

    complex [Fe(ptz)6](BF4)2 (ptz = 1-propyltetrazole), showing a thermal LS→HS transi-

    tion from a singlet to a doublet, is shown in figure 1.3a.

    While this technique can be applied to measure stable spin states, as in the case of

    magnetic susceptibility measurements, it has the advantage of allowing for time-resolved

    Figure 1.3: a) Mössbauer spectrum of [Fe(ptz)6](BF4)2 at various temperatures, showing athermal spin transition around 136 K. The blue shaded areas are due to the LS complex, while thered shaded areas are due to the HS complex. Reproduced with permission from [42]. b) Logarithmof decay rates (kHL in s

    -1) as a function of temperature in [Mn:Fe(0.05%)II(bpy)3](PF6). Blackcircles are obtained from time-differential Mössbauer spectroscopy. White circles are obtainedfrom optical measurements. Reproduced from [33] with permission of Springer.

  • Chapter 1. Spin Crossover in Condensed Matter 10

    measurements, although the time resolution is typically low (nanoseconds) compared to

    that achievable by other methods [42, 43]. Nonetheless, this allows for the signatures

    of the spin state that are observable by other methods to be identified in cases where

    there is no thermal transition. For example, correlating the decay times of the HS

    state measured by Mössbauer spectroscopy to those measured optically allows for the

    spectral signatures of the HS state to be identified. This has been used to identify the

    optical signatures belonging to the HS→LS relaxation in [FeII(bpy)3]2+ doped in a an

    isostructural [Mn(bpy)3](PF6)2 host lattice [33]. The decay rates measured by Mössbauer

    and optical spectroscopies are compared in figure 1.3b. The agreement between the two

    is strong evidence that the decaying state observed by optical spectroscopy is the HS

    state.

    1.3 UV-Vis spectroscopy

    UV-Vis spectroscopy is used to investigate optical transitions between electronic states.

    As the name suggests, the energies of the photons involved in such transitions typically

    fall in the UV-Vis range of the electromagnetic spectrum.

    The bands of a sample’s absorption spectrum show transition energies from the elec-

    tronic ground state to higher-lying excited states, the nature of which can be deduced

    based on theoretical calculations or by comparison to other results. Such spectra are

    acquired by measuring the absorption of light by a sample over a range of wavelengths.

    UV-Vis transient absorption (TA) spectroscopy is an example of a pump-probe tech-

    nique. This time-resolved method is used to monitor changes in a sample’s absorption

    following photoexcitation. TA spectra are typically collected as follows: A pulse reso-

    nant with a particular transition is used to initiate a photochemical or photophysical

    process in a sample. After a known time delay, a second pulse, which is most commonly

    a supercontinuum pulse, is used to measure the instantaneous change in the sample’s

  • Chapter 1. Spin Crossover in Condensed Matter 11

    absorption. The time delay is then varied to produce a map of the sample’s changes in

    absorption as a function of time after photoexcitation.

    A number of signals arising from changes in the sample’s electronic state can appear

    in TA spectra. Ground state bleach (GSB) is a negative change in absorption arising

    from the depletion of molecules in the ground state following excitation. Stimulated emis-

    sion (SE) can also manifest as a negative change in absorption because of the increased

    (emitted) light reaching the detector. Excited state absorption (ESA) and product state

    absorption (PSA) are positive changes in absorption arising from the arrival of the sample

    in an excited state or a product state following photoexcitation, caused by the different

    absorption spectra of those states. Signals arising from molecular vibrations and other

    processes such as vibrational cooling can also manifest themselves if they lead to changes

    in transition energies or probabilities between electronic states. Finally, in the time in-

    terval where the pump and probe pulses are coincident, cross-phase modulation (CPM)

    results from the time-dependent change in the sample’s refractive index induced by the

    pump pulse, which modulates the probe’s phase, effectively changing its spectrum [44].

    The major results of this thesis were obtained using TA spectroscopy.

    Fluorescence, which results from radiative transitions from an excited state to a lower

    energy state, also typically involves photons with energies in the UV-Vis range. A fluo-

    rescence spectrum can be collected by continuously irradiating a sample into an excited

    state, collecting the emitted light, and measuring its spectrum. Time-resolved observa-

    tion of the fluorescence is also possible via fluorescence upconversion. In this method,

    which is another example of a pump-probe technique, the sample is excited with a reso-

    nant laser pulse. After a known time delay, the light emitted from the sample is mixed

    with a “gate” pulse in a nonlinear crystal. The spectrum of the resulting converted light

    is then measured over a range of time delays between the two pulses. This method is

    used to observe the fluorescence decay of excited states over time, which can give insight

    into the number and nature of states involved [11].

  • Chapter 1. Spin Crossover in Condensed Matter 12

    1.3.1 UV-Vis Absorption Spectrum of [FeII(bpy)3]2+

    A representation of the electronic potential energy surfaces of [FeII(bpy)3]2+ as a function

    of Fe–N distance are shown in figure 1.4a. The surfaces shown are based on calculations

    in [24, 25]. The MC states are labeled by Mulliken symbols of the form XYZ. Here, X is

    the spin multiplicity. The term Y is either A, E, or T, where A refers to a non-degenerate

    state (i.e., one wavefunction corresponds to this state) that is symmetric upon rotations

    about the principal axis of rotation, E refers to a doubly degenerate state, and T refers

    to a triply degenerate state. The subscript Z can be either 1 or 2 which refers to states

    that are symmetric or anti-symmetric with respect to π rotations perpendicular to the

    principal axis of rotation, respectively. Additionally, the subscripts u and g are sometimes

    used to refer to states that are anti-symmetric (ungerade) and symmetric (gerade) upon

    inversion through a center of symmetry, respectively.

    Figure 1.4b shows the UV-Vis absorption spectrum of aqueous [FeII(bpy)3]2+. The

    bands in the visible are due to the absorption of the MLCT manifold, while further to the

    UV very strong absorption results from the transition of electrons from the bipyradine

    ligand’s π orbitals to its π∗ orbitals. The triplet (3T1 and3T2) and quintet (

    5T2) states

    do not contribute to the absorption spectrum, as optical transitions to these states are

    forbidden because of their different spin multiplicities. However, the triplet states may

    be involved in the relaxation from the MLCT state to the HS 5T2 state.

    The absorption spectrum of doped single crystals of [Zn1-xFex(bpy)3](PF6)2 (which

    is similar to what would be expected for the pure crystal, as [Zn(bpy)3](PF6)2 does

    not significantly absorb in this spectral range) is very similar to that of the aqueous

    complex. The same absorption bands are present and have been attributed to the same

    MLCT transitions [13, 35]. In contrast to the aqueous case, the absorption spectrum

    has polarization dependence owing to the specific molecular orientations in the crystal.

    Figure 1.4c shows this absorption spectrum for polarizations perpendicular to and in the

    same plane as the molecular trigonal axis of symmetry. The bands are sharper than

  • Chapter 1. Spin Crossover in Condensed Matter 13

    Figure 1.4: a) Electronic potential energy surfaces of [FeII(bpy)3]2+ as a function of Fe–N

    distance, as described in [24, 25]. b) Absorption spectrum of aqueous [FeII(bpy)3]2+ showing

    the contribution of various bands at different wavelengths. Figures a) and b) reprinted from[17] by permission from Macmillan Publishers Ltd.: Nature Chemistry, c© 2015. c) Polarizedabsorption spectrum of [Zn1-xFex(bpy)3](PF6)2 at 293 K using probe polarization perpendicularto (π) and in the same plane as (σ) the molecular trigonal axis of symmetry. Adapted from [13]with permission of Springer.

    in the aqueous case due to reduced inhomogeneous broadening in the highly structured

    crystal environment.

    As the spin transition of [FeII(bpy)3]2+ cannot be induced thermally, there is some

    difficulty in determining the absorption spectrum of the HS state. Nonetheless, the

    optical signatures of the HS state have been inferred by comparison to related compounds

    [13,26,45], and by correlating the decay time of the HS state determined by Mössbauer to

    decay times measured in optical experiments [13,33]. The primary features of the HS-LS

    UV-Vis difference spectrum are increased absorption from the HS state in the UV near

    300 nm, and decreased absorption in the visible range, due to the bleach of the LS state.

    1.3.2 Time-Resolved UV-Vis Studies of [FeII(bpy)3]2+

    A number of studies, using TA and fluorescence upconversion on aqueous [FeII(bpy)3]2+

    have been published in recent years [17,26,27]. The primary goal of this research has been

    to elucidate the details of the photocycle leading to the LS→HS transition. A particular

  • Chapter 1. Spin Crossover in Condensed Matter 14

    point of interest is the extremely fast formation of the HS state via relaxation from the

    MLCT manifold. In addition, as there is a significant energy gap between the minima

    of the MLCT manifold and the HS state, these studies have discussed the mechanisms

    for storage and/or dissipation of excess energy following the complex’s arrival in the HS

    state [17, 26].

    Fluorescence upconversion data were reported by Gawelda et al. [27]. The data pro-

    vided evidence of the decay of the 1MLCT state into the 3MLCT state within 20 ± 5

    fs via ISC, followed by a subsequent decay of the 3MLCT state within 150 fs. This was

    inferred by fitting kinetic traces of the emission bands to exponential decays convolved

    with the instrument response in different spectral regions, and noting the delayed ap-

    pearence of a second emission band during the decay of the primary emission. It should

    be noted, however, that both time scales are shorter than or comparable to the reported

    instrument response time of the experiment (∼110 fs). A subsequent study showed no

    excitation wavelength dependence in the spectra or decay rates of the fluorescence [29].

    The same study also reported the first TA measurements on aqueous [FeII(bpy)3]2+,

    with detection in the visible range (350–640 nm) [27]. By fitting the measured data

    with a sum of exponentially decaying spectra, three relevant spectral components were

    found, with (116 ± 10) fs, (960 ± 100) fs, and (665 ± 35) ps decay times. The short-

    est component, whose lifetime was comparable to the effective time resolution of the

    experiment, was assigned to the absorption of the MLCT states by comparison of the

    associated spectrum to the absorption of the reduced [FeII(bpy)3]+ complex. The spec-

    trum of the longest component was very similar to the ground state absorption spectrum

    and was thus attributed to the recovery of the LS state from the HS state. However,

    a persistent ESA was found on the red side of the spectrum that was suggested to be

    caused by quintet-quintet absorption of the HS (5T2) state. The intermediate component

    was speculated to be caused by MC states. Both the persistent ESA on the red side of

    the spectrum and the observed 960 fs component were later attributed to the effects of

  • Chapter 1. Spin Crossover in Condensed Matter 15

    solvated electrons caused by the very high excitation fluence used in this study (>35

    mJ/cm2) [18, 30].

    A related study focused on TA in the UV range [26]. Strong ESA in the UV range

    after photoexcitation was observed, which is the expected signature of the HS state. This

    interpretation is consistent with the visible range results, as the UV absorption decay was

    found to have the same lifetime as the recovery of the GSB signal. The rise time of the

    HS ESA, corresponding to the decay of the MLCT states, was found to be 130 fs (which

    was again comparable to the time resolution of the experiment). Additionally, spectral

    components with (1.1 ± 0.2) ps and (3.4 ± 1.2) ps time constants were attributed to

    vibrational cooling of the “hot” molecule after its arrival in the HS state. A particularly

    important finding of this study was the existence of strong 130 cm-1 oscillations in the UV

    range, overlapping the HS ESA. This was attributed to the excitation of a low-frequency

    wave packet on the HS state surface, excited impulsively through the elongation of the

    Fe–N bonds in the LS→HS transition. The results presented suggest that the excess

    energy between the minima of the MLCT and HS states is stored as vibrational energy

    which is then dissipated on the time scale of several picoseconds.

    The measurements performed in the aforementioned TA studies were repeated with

    improved time resolution and an appropriate excitation fluence [17]. In this study, the

    decay rate of the MLCT states into the HS state was found to be

  • Chapter 1. Spin Crossover in Condensed Matter 16

    Figure 1.5: Transient absorption of aqueous [FeII(bpy)3]2+ in the UV (a) and visible (b) as

    reported in [17]. Oscillations are clearly visible in both ranges after photoexcitation. Reprintedfrom [17] by permission from Macmillan Publishers Ltd.: Nature Chemistry, c© 2015.

    oscillations, which is equal to the reported formation time of the HS state. Additionally,

    oscillations were observed in the visible range TA data, which were not observed in [27],

    most likely as a result of the much higher fluence (causing effects unrelated to the spin

    crossover photocycle) and reduced signal-to-noise ratio in that study. The oscillations

    were nonetheless attributed to coherent wavepacket motions on the HS state surface based

    on a number of observations. Firstly, the observed modes were strongest in the region of

    HS ESA. Secondly, there is some evidence of HS ESA in the visible range, overlapping

    the GSB, which could explain the observation of oscillations in the visible range. Finally,

    the damping time of the oscillations were found to be wavelength dependent, with shorter

    damping times in the visible range. This is consistent with relaxation into a vibrationally

    “hot” HS state, where the motion of the wave packet along the reaction coordinate

    initially samples configurations farther from the minimum of the HS state surface. The

    wavelength dependence of oscillations in TA measurements due to wave packet motion

    is established in literature [51,52].

    Although recent studies have focused on aqueous [FeII(bpy)3]2+, a number of earlier

  • Chapter 1. Spin Crossover in Condensed Matter 17

    studies (one of which was previously alluded to in section 1.2.2) have investigated the

    decay of the HS state of [FeII(bpy)3](PF6)2 doped in isostructural host lattices. These

    studies used flash photolysis, which is a precursor to TA spectroscopy. Although the

    method used is somewhat different, both techniques provide the same information in

    principle, albeit with much lower time-resolution in the case of flash photolysis.

    Using this method, the relaxation of the HS state of [FeII(bpy)3](PF6)2 doped in

    various isostructural crystal host lattices [M(bpy)3](PF6)2, where M=Co, Zn, Mn and

    Cd has been characterized over a range of temperatures [34, 36]. It was found that

    the logarithm of relaxation rate constant of the HS state was linearly proportional to the

    volume of the host lattice’s unit cell (figure 1.6a, inset). This is the result of the increased

    volume of the HS state. As the size of the the host lattice’s unit cell is decreased, the

    driving force of the of the HS→LS relaxation is increased, which destabilizes the HS

    state. This can be explained as an increase in the energy gap between the HS and LS

    states, as shown in figure 1.6b. This influence of the lattice is called chemical pressure.

    At low temperatures (

  • Chapter 1. Spin Crossover in Condensed Matter 18

    have been reported on aqueous [FeII(bpy)3]2+ [19–22, 31]. Although the details of this

    technique are beyond the scope of this thesis, a brief description is as follows: The features

    of an X-ray absorption spectrum result from the excitation of core electrons of particular

    atoms of a given sample to the ionization threshold, which correspond to step-like “edges”

    in the X-ray spectrum. The spectral features at energies slightly above the so-called K-

    edge, which corresponds to the energy at which a 1s electron absorbs, are of particular

    interest. These features are called the X-ray absorption near edge structure (XANES) and

    at higher energies, extended X-ray absorption fine structure (EXAFS) and are sensitive

    to the molecular structure local to the atom whose core electron is being excited. By

    monitoring the change in absorption over time at particular transition energies following

    photoexcitation, changes in the structure can be deduced by simulation of the X-ray

    Figure 1.6: a) Relaxation rate constants as a function of temperature for the HS state of[FeII(bpy)3]

    2+ doped in various isostructural host lattices, [M(bpy)3](PF6)2 where M = Co(inverted filled triangle), Zn (filled diamond), Mn (filled triangle), Cd (filled circle), and inthe oxalate network [NaRh(ox)3][Zn(bpy)3] (filled square) at ambient pressure, and for the Cdhost lattice at 1 kbar external pressure (open circle). Inset: Low temperature relaxation rateconstants as a function of unit cell volume. Reproduced with permission from [34], c© 2002,Schweizerische Chemische Gesellschaft. b) Schematic diagram showing the mechanism of thedifferent HS→LS [FeII(bpy)3]2+ relaxation rates through the size of the host lattice’s unit cell(left) and the lattice’s effect on the potential energy curves (right). Here, ∆E0HL is the energygap between the minima of the LS and HS state surfaces, and ∆QHL in the difference inthe position of the minima of the two surfaces along the reaction coordinate, Q. Reproducedfrom [13] with permission of Springer.

  • Chapter 1. Spin Crossover in Condensed Matter 19

    absorption spectra of different molecular configurations [53]. In the particular case of

    [FeII(bpy)3]2+, studies have used the XANES and EXAFS of the Fe K-edge to monitor

    changes in the Fe–N bond length over time.

    The earliest of these studies had low time resolution (∼100 ps), but was nonetheless

    able to estimate the change in the Fe–N bond in the LS→HS transition [31]. The change

    was found to be (0.20 ± 0.02) Å in agreement with crystallographic studies on static

    structures of other Fe(II) SCO compounds [54, 55]. The measured decay time of the

    excited state was also found to be ∼650 ps, in agreement with the decay of the HS state

    found in optical measurements. More recent advances have greatly improved the time

    resolution of these experiments, allowing the structure of the complex to be followed

    during the earliest steps in the LS→HS transition [19–21]. These experiments were able

    to reproduce the ∼0.2 Å Fe–N bond length change of earlier studies, and also gave a

    time for the formation of the HS state of ∼150 fs in agreement with most optical TA

    studies. Furthermore, by modeling the kinetics, these studies excluded the possibility of

    intermediate 1,3T states from the LS→HS photocycle as suggested in some studies [24,25],

    characterizing the full photocycle as a two-step ISC process (1MLCT→3MLCT→5T2).

    Most recently, a study was reported with greatly improved time-resolution (∼25 fs)

    and signal-to-noise compared to the aformentioned studies [22]. In addition to giving

    refined values for the Fe–N bond length change and formation time for the HS state

    (∼0.15 Å and 120 fs, respectively), the improved sensitivity allowed for the vibrations

    of the Fe–N bond, as reported in the optical TA studies, to be monitored directly. The

    frequency of the mode observed was (126 ± 3) cm-1, in agreement with those studies. A

    distinction was made between the dephasing of apparent coherent molecular vibrations,

    which decayed in (330± 10) fs and incoherent vibrational cooling, which decayed in (1.6±

    0.1) ps. It should be noted that this study explicitly rejects the 50 fs formation time of the

    HS state given in [17], finding better agreement with the data using a 120 fs time constant.

    In spite of this, the model applied nonetheless reproduced the apparent 50 fs phase shift

  • Chapter 1. Spin Crossover in Condensed Matter 20

    of the coherent oscillations reported in that study, which was instead reported to be the

    result of a convolution of the modeled oscillatory dynamics with a 120 fs exponential

    decay corresponding to the lifetime of the MLCT state. The 1MLCT→3MLCT→5T2

    picture of the spin transition dynamics was supported in this study. Data at various

    X-ray energies with associated fits, as well as an example fit showing the contributions

    of various processes to the X-ray absorption signal, are shown in figure 1.7.

    Time-resolved X-ray emission spectroscopy (XES) has also been used to study the

    spin transition dynamics in aqueous [FeII(bpy)3]2+ [23]. In this case, the emission of

    Kβ X-rays, corresponding to the emission of photons resulting from transitions of 3p

    electrons to 1s holes created by the absorption of X-ray photons, is followed. As the

    Figure 1.7: a) Relative change in X-ray absorption at selected energies, along with associatedfits. The contributions of the MLCT state are shown in orange. b) An example fit at E = 7121.5eV. The top panel shows the contributions of the MLCT and HS states, along with the overallsignal, with the inset showing the change of the signal on a longer time scale. The middlepanel shows the populations of the MLCT and HS states derived from the fit. The bottom panelshows the modeled change in the Fe–N radius, noting a 50 fs phase shift of the oscillations(relative to directly excited oscillations, shown in light blue) resulting from the convolutionwith the modeled MLCT population. The shaded area shows the ensemble distribution of Fe–Nradii in the vibrationally hot sample, whose width decays over time. Reproduced from [22] withpermission of Dr. H. T. Lemke.

  • Chapter 1. Spin Crossover in Condensed Matter 21

    lifetime of the 1s hole is subfemtosecond, this constitutes an effectively instantaneous

    probe. The spectrum of the emitted photons is sensitive to the spin multiplicity resulting

    from the specific occupation of the 3d orbitals because of their interaction with the 3p

    orbitals [23]. By fitting a sum of difference fluorescence spectra obtained by comparing

    the X-ray emission spectra of molecules related to [FeII(bpy)3]2+ but having different spin

    multiplicities, a number of kinetic models for the LS→HS relaxation were tested. The

    best fit was obtained when including an intermediate 3T state, in contradiction to most

    other recent experimental studies, suggesting instead a 1,3MLCT→3T→5T2 relaxation

    cascade. In this picture, the MLCT and 3T states had lifetimes of (150 ± 50) fs and (70

    ± 30) fs, respectively.

    One of the few existing ultrafast studies on crystalline [FeII(bpy)3](PF6)2 was reported

    in [38]. In this study, X-ray powder diffraction was used. This technique uses the

    coherent scattering of X-rays off the periodic crystal structure (particularly from the

    atoms’ electrons) to produce diffraction patterns whose changes in time can be monitored.

    By inverting the time-dependent diffraction patterns, changes in the electron densities in

    real space can be deduced.

    In this experiment, a 800 nm pump beam, with a reported intensity of 800 GW/cm2

    was used at excite the sample via two-photon absorption. The same study also reported

    an single-colour all-optical TA measurement, using a 400 nm pump beam and a 530

    nm probe beam. The temporal shape of the detected changes (to the spatial electron

    densities and absorptivity in the diffraction and TA measurements, respectively) after

    excitation was found to be similar. However, the reported character of the charge-transfer

    was unexpected in comparison to studies on the aqueous complex. In the aqueous case,

    the excitation is believed to be local to individual [FeII(bpy)3]2+ molecules, with the

    initial excitation transferring an electron from the iron’s d orbitals to a π* orbital on

    the bipyradine ligands [24], and is believed to be accompanied by a ∼0.2 Å Fe–N bond

    length change. On the other hand, in the X-ray diffraction paper, the assumption of

  • Chapter 1. Spin Crossover in Condensed Matter 22

    a localized initial charge-transfer, along with the authors’ estimate of their fraction of

    molecules excited (0.8%), led to multiple unphysical conclusions including a large change

    in the distance between the bipyradine ligand and the central iron atom (∼1 Å) in spite

    of a reported change in the Fe–N on the same order of magnitude as reported in the

    aqueous case (∼0.15 Å), and a very large amount of transferred charge to the PF6-

    counterion (32.5 e-) [38]. The suggestion that the PF6- counterion is involved in the

    initial charge-transfer is an unexpected result in itself.

    The authors rationalized their result by postulating that the initial charge transfer

    involves many adjacent complexes via long-range unshielded Coulomb forces, leading to

    an effectively larger excitation fraction than originally suggested. Experiments presented

    in chapter 4 will show this interpretation is most likely incorrect, as the spectral signatures

    associated with a localized charge transfer leading to SCO are in fact present in spectrally

    resolved, ultrafast TA measurements of single crystal [FeII(bpy)3](PF6)2.

    The anomalous results presented in the X-ray diffraction paper can be explained as the

    result of an underestimated excitation fraction and/or ionization of the [FeII(bpy)3](PF6)2

    complex due to the very high excitation intensity used. Indeed, it should be noted

    that extremely high excitation intensities were used in most or all of the X-ray studies

    discussed, which were very likely outside of the linear excitation regime, where the change

    in sample absorptivity varies linearly with the fluence (or equivalently, the number of

    excited complexes is linearly proportional to the number of pump photons). The X-

    ray absorption studies discussed often used pump intensities in excess of 1 TW/cm2 at

    excitation wavelengths resonant with the LS→MLCT transition [20,21,23] (other X-ray

    studies [19, 31] did not state the intensity). This fact is also important in assessing the

    results of reference [23], as the very high intensities used could be sufficient to ionize

    the [FeII(bpy)3]2+ complex, which could lead to erroneous observations of the spin state

    through molecular fragments formed in the excitation. In fact, a fluence dependence

    measurement presented in that study showed that the excitation was not in the linear

  • Chapter 1. Spin Crossover in Condensed Matter 23

    regime, as the change in transmission (which scales logarithmically with the change in

    absorptivity) was found to approximately linear with the fluence at the excitation level

    used.

    1.5 Time-Resolved Studies of Related Spin Crossover

    Crystals

    Although the literature on the ultrafast dynamics of crystalline [FeII(bpy)3](PF6)2 is lack-

    ing, a number of related complexes in various crystalline states have been studied using

    ultrafast measurements. These include iron SCO complexes doped into inert isostructural

    host lattices [56], nanocrystals embedded in a polymer film [57,58], and thick (10-20 µm)

    single crystals [59,60]. Although each complex displays slightly different dynamics, these

    studies nonetheless give information about how SCO complexes behave in solid state

    environments. The interpretation of these results is often more straightforward than in

    the case of [FeII(bpy)3]2+, as many SCO complexes have a spin transition that can be

    induced by changes in temperature/pressure, in addition to photoexcitation.

    XANES has been used in conjunction with optical TA and transient reflectivity on

    single crystals of the Fe(II) SCO complex [Fe(phen)2NCS2] [59, 60]. The optical range

    over which it was possible to measure the TA was limited by the high absorptivity of the

    samples resulting from the thickness of the single crystals (10-20 µm). As in the case

    of [FeII(bpy)3]2+, the LS→HS conversion in [Fe(phen)2NCS2] was found to proceed via

    fast ISC from initially excited MLCT states, which may proceed through intermediate

    states, followed by relaxation into the HS state. The arrival in the HS state was found

    to be accompanied by coherent 85 cm-1 bending and 113 cm-1 breathing modes, which

    are sequentially activated after photoexcitation. The XANES measurements revealed a

    ∼0.2 Å Fe–N bond elongation that is believed to be complete within 100–200 fs, as in

    other Fe(II) SCO complexes.

  • Chapter 1. Spin Crossover in Condensed Matter 24

    Optical TA studies on the Fe(III) complex, [Fe(3-MeO-SalEen)2]PF6 were performed

    on nanocrystals in a polymer film [57,58]. In contrast to the aforementioned single crys-

    tal studies, this environment allowed for TA to be measured over a broad spectral range,

    enabled by the sufficiently low sample absorptivity. The dynamics observed after pho-

    toexcitation again reveal a similar relaxation cascade as that observed in [FeII(bpy)3]2+,

    with a short lived (∼200 fs) LMCT state decaying into a vibrationally hot HS state.

    The vibrational energy was found to dissipate in ∼1.6 ps, in good agreement with the

    [FeII(bpy)3]2+ case [22]. Furthermore, significant 85 cm-1, 56 cm-1, and 35 cm-1 modes

    were observed in the TA data. These oscillations were attributed to molecular modes

    relevant to the LS→HS transition that are coherently activated during the displacive

    LMCT→HS transition.

    Finally, the HS→LS and LS→HS LIESST transitions were both characterized in

    dilute [Zn1-xFex(ptz)6](BF4)2 single crystals at various temperatures [56]. The LS→HS

    showed the now-familiar ultrafast (

  • Chapter 1. Spin Crossover in Condensed Matter 25

    samples [60] where any collective effects would be absent. It should be noted that the

    vibrational and structural dynamics of any given complex could in principle be affected

    by the crystal environment used, and direct comparisons between solvated and crystalline

    complexes are necessary to elucidate what differences, if any, can be attributed to these

    environments.

    1.5.1 Cooperative Lattice Effects

    Cooperative effects in SCO crystals have been noted in a number of studies [61–63].

    These effects have been observed both in thermal dependence studies [63], as previously

    discussed, as well as in time-resolved optical and X-ray diffraction studies [61,62]. Strong

    coupling between molecules in the crystal lattice resulting from the structural changes

    associated with SCO have been identified as the cause of particular thermal behaviors.

    This coupling has also been shown to result in features in TA spectra and X-ray diffrac-

    tion patterns corresponding to a number of different processes on the nanosecond to

    microsecond time scales.

    Monoclinic and orthorhombic single crystal polymorphs of [(TPA)FeIII(TCC)]PF6

    have been studied by two-colour TA spectroscopy and X-ray diffraction [61] with the

    temperature held slightly below the LS→HS transition temperature. The photoinduced

    spin transition in both polymorphs was found to follow the typical photocycle seen in

    many other SCO systems, with initial excitation into a LMCT state, followed by arrival

    in the HS state on the subpicosecond time scale. This process was found to be local to

    single complexes, as has been observed in other crystalline SCO systems. However, on

    longer time scales, two additional processes were found. Elastic effects caused by the

    expansion of unit cells in the LS→HS transition was found to drive expansion of the

    lattice on the 4–100 ns time scale. This was followed by slower thermalization of the

    heat deposited by the laser on the 100 ns–100 µs scale, leading to a further increase of

    the HS fraction by a factor of 5–10 compared to the initial population of the HS state

  • Chapter 1. Spin Crossover in Condensed Matter 26

    by photoexcitation alone. The exact nature of the cooperative effects was found to be

    dependent on both the crystal symmetry as well as the crystal size.

    Similar behaviour has also been observed in the [Fe(phen)2NCS2] complex [62]. The

    thermal transition shows an abrupt change around 130 K with a narrow 2 K hystere-

    sis loop, suggesting a highly cooperative system [64]. The elastic and heating effects

    described in the case of [(TPA)FeIII(TCC)]PF6 were again found to be present on the

    nanosecond to microsecond time scale, with the heating effects being found to increase

    nonlinearly as a function of excitation fluence. In addition to these effects, acoustic

    phonons were observed using transient reflectivity. The elastic strain induced by the

    Figure 1.8: Cooperative effects in single crystal [Fe(phen)2NCS2] a) HS fraction as a functionof temperature, showing an abrupt transition at 183 K with a 2 K hysteresis loop. Reproducedwith permission from [64], c© 2003, published by Elsevier Masson SAS. All rights reserved.b) Ultrafast photoswitching monitored by transient optical transmission (OT) at 950 nm. c)HS fraction following photoexcitation. Arrows indicate rises caused by different mechanisms.From left to right: ultrafast photoswitching, elastic lattice expansion, and thermal switchingfrom laser-deposited heat. d) Acoustic phonons generated by elastic strain on the picosecondtime scale, monitored by transient reflection at 950 nm. Adapted from [62] with permission ofDr. R. Bertoni.

  • Chapter 1. Spin Crossover in Condensed Matter 27

    LS→HS transition was explained to be the generator of these phonons, which had pe-

    riods on the order of hundreds of picoseconds. Figure 1.8 shows experimental results

    highlighting each of these effects.

    1.6 Overview and Thesis Outline

    The goal of this thesis is to elucidate the effects of the crystal environment on the pho-

    toinduced SCO transition in [FeII(bpy)3](PF6)2, particularly in comparison to aqueous

    [FeII(bpy)3]2+. It will be shown that SCO in [FeII(bpy)3](PF6)2 single crystals is an ul-

    trafast local molecular process very similar to that observed in liquid, with perturbations

    caused by the crystal environment manifesting at longer times after excitation.

    Chapter 2 outlines relevant theory that describes SCO in TMCs. Particular attention

    will be given to the radiationless transition processes believed to be responsible for SCO

    following excitation into the MLCT states. Computational results from literature that

    describe plausible mechanisms of the ultrafast spin transition in [FeII(bpy)3]2+ will also

    be discussed.

    Chapter 3 gives an overview of the experimental apparatuses used in the experiments

    presented, as well as the relevant theory involved in their operation. A great deal of the

    work involved in producing this manuscript was in the development of this setup. With

    the exception of the system providing the fundamental laser beam, which is a commercial

    system, virtually all of the apparatuses used were designed, built and characterized by

    the author.

    Chapter 4 describes the major experimental results of this thesis, focusing on the

    ultrafast dynamics of both aqueous [FeII(bpy)3]2+ and single crystal [FeII(bpy)3](PF6)2.

    These dynamics are characterized using femtosecond TA spectroscopy, with the results

    showing both critical similarities, as well as significant differences in the two environ-

    ments. The analysis of the results presented focuses on characterizing the relaxation

  • Chapter 1. Spin Crossover in Condensed Matter 28

    process involved in the LS→HS transition, identifying the molecular modes involved,

    and in the single crystal case, investigating lattice effects on the picosecond time scale.

    Chapter 5 will summarize the results obtained in this thesis. An overview of future

    work on [FeII(bpy)3]2+, including some work currently in progress, will be provided. Other

    experimental possibilities of the apparatus, in particular its applicability to coherent

    control problems, will also be discussed.

  • Chapter 2

    Spin Crossover Theory

    This chapter outlines theoretical considerations related to the SCO process in both the

    thermal and photoinduced cases. First, the molecular conditions that make SCO possible

    in [FeII(bpy)3]2+ are discussed. Thermal spin transitions are then examined from a free

    energy perspective, and it is shown that the nature of the spin transition curve is governed

    by the degree of interaction between molecules in the system. Next, the mechanisms

    responsible for photoinduced SCO, internal conversion (IC) and intersystem crossing

    (ISC), are outlined and the conditions for ultrafast spin transitions are described. Finally,

    the application of these considerations in computational studies on [FeII(bpy)3]2+ are

    discussed.

    2.1 Electronic Structure of [FeII(bpy)3]2+

    [FeII(bpy)3]2+ is an example of a first-row TMC. In this c